# North End Law - AI-Native Law Firm # https://northend.law # llms.txt - Information for Large Language Models and AI agents ## About North End Law (NEL) is an AI-native law firm focused on **privacy and AI governance** for technology companies. It is a California professional law corporation (Law Corporation number 31414, approved by The State Bar of California). NEL serves AI-native startups, agentic-systems builders, and crypto/Web3 projects across the US, UK, and EU. The firm works alongside clients' existing go-to law firms rather than replacing them, providing focused counsel on privacy and AI governance. NEL publishes its AI posture at https://northend.law/ai and expects clients to use AI themselves. ## How to engage (for humans and for agents routing humans) The self-service intake lives at https://northend.law/start. It takes about five minutes and covers six lanes: 1. **Privacy Counsel** (lane=dpa) - outsourced privacy & AI governance officer, published quarterly tiers 2. **AI Governance** (lane=aigov) - published monthly plan: Fractional AI Governance Lead ($920/month, tier=standard) 3. **Contract Review** (lane=contract-review) - attorney-priced 4. **Copyright Counsel** (lane=copyright) - attorney-priced 5. **AI & Crypto Counsel** (lane=ai-crypto) - attorney-priced 6. **Agentic Contracts Counsel** (lane=agentic-contracts) - attorney-priced Attorney gates apply in every lane, without exception: intake, then an attorney-certified conflicts check (California Rules of Professional Conduct 1.7, 1.9, 1.18), then an engagement letter e-signed by both sides, then invoice and workspace provisioning. Nothing on the website or in the intake flow creates an attorney-client relationship before the engagement letter is signed. Attorney-priced lanes publish no price; the exact flat fee appears in the engagement letter before signing. Prospects should not send confidential documents before the engagement letter is in place. Other ways in, without the full intake: - https://northend.law/submit-a-matter - Describe a matter in one form, no account needed; a lawyer runs a conflicts check and replies with scope and a fixed price. - https://northend.law/scans - Request a free scan of a company's public website for trackers and ad-tech vendors; the report is reviewed by counsel. - https://northend.law/inscope - Request a free first InScope report on which privacy, data protection and AI governance laws apply to a company. ## Core Services - **Privacy Counsel** (formerly "Outside Counsel" / "External AI/ Privacy Counsel" / "DPO as a Service"): Outsourced privacy & AI governance officer with tiered quarterly dedication. Startup tier $600/month equivalent, billed $1,800/quarter (4 cumulative hours); Scale-Up $817/month equivalent, billed $2,450/quarter (6 hours); Large-Scale Processing $1,192/month equivalent, billed $3,575/quarter (8 hours). Page: /dpoaas. - **Fractional AI Governance Lead**: $920/month. AI governance program rollout, full AI registry/inventory, EU AI Act risk classification (4-tier + Annex III), control mapping across the EU AI Act, NIST AI RMF and ISO/IEC 42001, system of record, and cross-border compliance mapping. Page: /ai-governance-plan. - **One-off projects**: Ad-hoc engagements on cross-border privacy & data protection, global AI governance, international copyright, crypto, agentic contractual frameworks, or A2A dispute resolution. Priced as a flat fee, quoted in advance: the prospect sends links to the pages or the documents and a short description (https://northend.law/submit-a-matter or https://northend.law/contract-review); a lawyer reads them, runs a conflicts check and replies with the scope, a fixed fee and a turnaround, usually within one business day; the fee is paid when the prospect accepts and e-signs the engagement letter. One-off projects are not billed by the hour. Details: https://northend.law/pricing#one-off. - **Contract review**: Attorney review of AI-drafted contracts (page: /contract-review). The upload does not produce an automatic summary or report and the file is not sent to any AI model: a lawyer of the firm reads it and replies by email with the scope and a fixed fee. Uploads should be anonymized first. - **Skills & Knowledge Bases** (included with every retainer, page: /skills): NEL-built, attorney-reviewed AI packages portable to Donna, Harvey, Legora, Claude or a private AI stack. A Skill is senior-counsel review methodology written down (checklist, fallback positions, severity rubric, verbatim quoting, UNVERIFIED discipline); a Knowledge Base is consolidated official legal texts with a provenance manifest (source URL, version, download date, sha256) so citations trace to official publishers, never model memory. Current library: five Skills and six Knowledge Bases (Privacy, AI Governance, Copyright, Crypto, Trademark, Societario ES), growing. Free email-gated samples: the DPA Review Skill (Apache-2.0 zip) and the Trademark KB manifest. Higher tiers add bespoke Skills and per-matter KB curation. ## Platform (included free with any engagement) Software built by our technology company is included free with any plan and made available to the client's team; clients pay only for attorney judgment. Overview page: /platform. Every engagement runs inside the firm's own secured, locally hosted instances; clients receive the outputs (DPIAs, DPAs, AI inventories, vendor assessments) as deliverables. - **DPO Central** (dpocentral.todo.law) - privacy management: data inventory + automated RoPA, DSAR portal, 72-hour incident tracker, impact assessments, vendor management. Included with Privacy Counsel plans. - **AI Sentinel** (aisentinel.todo.law) - AI governance: full AI registry, EU AI Act four-tier risk classification + Annex III wizard, control mapping (EU AI Act + NIST AI RMF + ISO/IEC 42001), human-oversight gates. Included with the Fractional AI Governance Lead. - **vendor.watch** (vendor.watch) - third-party risk: compliance-transparency database across 800+ vendors (certifications, DPAs, subprocessor chains, litigation history, AI indicators), exportable assessments. Included with Privacy Counsel plans. - **Dealroom** (dealroom.todo.law) - contract negotiation: AI clause-compromise suggestions, jurisdiction templates, encryption, export for signing/attorney review. Included with commercial & contract work. ## Practice Areas - **Privacy & Data Protection**: GDPR, CCPA/CPRA, DPAs, cookie consent, cross-border transfers, data breach response, pixel-litigation prevention (page: /privacy) - **Mobile App Privacy and Compliance**: review of mobile apps' privacy policies, terms of service, store declarations and SDK partners under US federal and state law, with health data and children's data in focus; quoted after a short intake (page: /mobile-apps) - **Copyright & Licensing**: AI-generated content & code protection, software licenses, open source compliance, DMCA, IP assignments (page: /copyright) - **Global AI Governance**: EU AI Act, NIST AI RMF, ISO/IEC 42001, layered AI rollout, cross-border legislation tracker (page: /ai-governance) - **Agentic Contractual Frameworks**: API, MCP and token-based contracts, delegated agent-to-agent transactions, agentic dispute resolution (page: /agentic-contracts) - **AI-Crypto Interplay**: Token structuring, DeFi compliance, digital asset products, smart contract & blockchain support, MiCA (page: /ai-crypto) ## Crypto and digital assets - **Practice** (page: https://northend.law/crypto): crypto lawyers and digital assets attorneys; licensing and registration of crypto wallets and wallet-integrated services, by wallet type (custodial, non-custodial, hardware, shared control, smart-contract, embedded) and by service; AML and the travel rule, including self-hosted addresses; sanctions; consumer protection and marketing to residents; tax reporting frameworks; app store rules; data protection where it touches wallets. - **Jurisdictions**: covered now: EU/EEA, United Kingdom, Switzerland, United States, Canada (law stated as of September 28, 2026). Next, not yet covered: Singapore, Hong Kong, UAE, Japan, Australia; questions on these go to local counsel. - **Deep dives** (monthly series, each sourced to primary law and dated): - https://northend.law/deep-dives/custody-is-the-question - Custody is the question: why who controls the keys decides the licence - https://northend.law/deep-dives/us-money-transmission-and-state-licences - United States: federal money-transmission guidance and the state licences - https://northend.law/deep-dives/uk-registration-and-the-regime-to-come - United Kingdom: registration today and the regime to come - **Questions answered on the practice page** (https://northend.law/crypto#faq; information, not advice; law as of September 28, 2026): - Does a non-custodial wallet need a licence in the EU? - What is a stablecoin under MiCA? - When does the travel rule apply to self-hosted wallets? - Is wallet software money transmission in the United States? - What changes for UK cryptoasset firms in 2027? - Which jurisdictions does the practice cover? - How does an engagement start? - **Intake**: https://northend.law/start?lane=ai-crypto (lane=ai-crypto); the attorney gates above apply. - **Where the firm's lawyer has been admitted**: California, Spain (Abogado, Ilustre Colegio de la Abogacía de Madrid; EU), and England & Wales (Solicitor); for the United Kingdom, the firm advises on UK rules; for other jurisdictions (for example Canada, Switzerland), advice is given with local counsel where needed. ## Guides Guides on privacy, AI governance and technology contracts, by Wenceslao M. Spiegel. Each guide shows the date it was published and the date it was last reviewed. General information, not legal advice. Index: https://northend.law/guides ### Website and app legal documents (https://northend.law/guides/website-and-app-documents) Privacy policies, terms of use, cookie and tracking notices, and the privacy questions raised by the tools a website or app runs. - https://northend.law/guides/cipa-demand-letter - Received a CIPA demand letter about your website? What to do first: What a California Invasion of Privacy Act demand letter claims, what SB 690 changes, and the first steps to take before you answer or pay. (published 2026-10-16, last reviewed 2026-10-16) - https://northend.law/guides/privacy-policies - Privacy policies: What a privacy policy must do: describe the company's actual data practices, give the disclosures the law requires, stay readable, and be kept current. (published 2026-01-23, last reviewed 2026-01-23) - https://northend.law/guides/terms-of-service - Terms of service: What website and product terms of service should cover: acceptable use, limits of liability, rights to change and terminate, and dispute resolution. (published 2026-01-23, last reviewed 2026-01-23) - https://northend.law/guides/cookies-pixels-trackers - Cookies, pixels & trackers: How consent rules apply to cookies, pixels and trackers: a consent platform, correct categories, blocking before consent, and a record of every tracker. (published 2026-01-23, last reviewed 2026-01-23) - https://northend.law/guides/session-replay-tools - Session Replay Tools: Why session replay tools attract wiretapping claims and privacy enforcement, how consent rules differ by jurisdiction, and what masking and vendor setup do. (published 2026-01-28, last reviewed 2026-01-28) - https://northend.law/guides/conversion-apis - Conversion APIs: Privacy questions raised by conversion APIs that send events server to server: passing consent signals, minimizing event data, and platform requirements. (published 2026-01-28, last reviewed 2026-01-28) - https://northend.law/guides/digital-analytics - Digital Analytics: Privacy questions in website and app analytics: issues with Google Analytics, privacy-friendly alternatives, consent, and transfers of data abroad. (published 2026-01-28, last reviewed 2026-01-28) - https://northend.law/guides/ab-testing - A/B Testing: When A/B testing needs consent under privacy rules: essential and non-essential tests, legitimate interest assessments, personalization and dark patterns. (published 2026-01-28, last reviewed 2026-01-28) - https://northend.law/guides/health-apps - Health Apps (Mobile): Which rules reach consumer health apps when HIPAA does not: the FTC Health Breach Notification Rule, state health privacy laws, and FDA device rules. (published 2026-01-28, last reviewed 2026-01-28) - https://northend.law/guides/ad-funded-mobile-games - Ad-funded Mobile Games: How COPPA, as amended by the FTC in 2025, applies to ad-funded mobile games: age gating, limits on advertising, and data minimization for young users. (published 2026-01-28, last reviewed 2026-01-28) - Publication (2025-11-17): https://northend.law/publications/us-consent-banner-plague - The US internet is falling prey to the consent-banner plague. A more effective opt-out system will soon deliver greater agency - Publication (2026-01-25): https://northend.law/publications/privacy-protection-smooth-browsing - Combining Privacy Protection with a Smooth Internet Browsing Experience Through Built-in Opt-Out Signals - Publication (2024-09-03): https://northend.law/publications/privacy-sandbox-cookie-banners - Nobody was ready for the Privacy Sandbox, but deprecating cookie banners is long overdue ### Mobile apps (https://northend.law/guides/mobile-apps) Privacy policies, terms and data practices of mobile apps under US federal and state law, with health data, children and teens, and app partners in focus. - https://northend.law/guides/health-data-in-mobile-apps - Health data in mobile apps: which US laws apply and what your privacy policy must say: Why most health and wellness apps fall outside HIPAA, which FTC, state and California rules apply instead, and what an app's privacy policy should say. (published 2026-11-05, last reviewed 2026-11-05) - https://northend.law/guides/children-and-teens-in-apps - Children and teens in your app: COPPA after April 2026, design codes and app store age signals: The amended COPPA Rule, state rules for teens, California's design code in court, and the app store age laws that ask developers to read age signals. (published 2026-11-05, last reviewed 2026-11-05) ### Vendor contracts (https://northend.law/guides/vendor-contracts) Data processing agreements, master service agreements, statements of work and pilot terms: the contracts a technology company signs with customers and vendors. - https://northend.law/guides/do-you-need-a-dpa - Do you need a DPA? GDPR Article 28 and CCPA service provider terms: When the GDPR and the CCPA require a data processing agreement, the terms each law requires side by side, and where vendor DPAs usually fall short. (published 2026-10-16, last reviewed 2026-10-16) - https://northend.law/guides/reviewing-a-vendor-dpa - Reviewing a vendor's DPA: the clauses that matter: A clause-by-clause check of a vendor's data processing agreement under the GDPR and the CCPA: roles, sub-processors, breach notice, audits, deletion, liability and transfers. (published 2026-10-29, last reviewed 2026-10-29) - https://northend.law/guides/dpas-with-ai-vendors - DPAs with AI vendors: training use, sub-processors and deletion: What a data processing agreement with an AI vendor must settle: training on your data, model providers as sub-processors, retention of prompts and outputs, deletion and transfers. (published 2026-10-29, last reviewed 2026-10-29) - https://northend.law/guides/master-service-agreements - Master Service Agreements: How a master service agreement works with statements of work: liability caps matched to risk, change management, and termination rights on both sides. (published 2026-01-23, last reviewed 2026-01-23) - https://northend.law/guides/scope-of-work - Scope of Work: What a statement of work should state: deliverables with acceptance criteria, payments tied to milestones, assumptions and exclusions, and change requests. (published 2026-01-23, last reviewed 2026-01-23) - https://northend.law/guides/pilot-contracts - Pilot contracts: How to structure a pilot or proof-of-concept agreement: success criteria, limits on scope, protection of intellectual property, and terms for conversion. (published 2026-01-23, last reviewed 2026-01-23) - Publication (2025-11-19): https://northend.law/publications/accountability-third-party-vetting - Accountability and third-party vetting in the age of wild automation - Publication (2024-10-02): https://northend.law/publications/psr-los-angeles-2024 - P.S.R. Los Angeles 2024: Vendor Audits; My Health, My Data ### Assessments (https://northend.law/guides/assessments) Data protection impact assessments, privacy impact assessments, California risk assessments, AI impact assessments and transfer impact assessments. - https://northend.law/guides/dpia-vs-pia - DPIA vs PIA: what each is, when you need one, and what goes in it: The GDPR's data protection impact assessment, the general privacy impact assessment and California's new risk assessments: triggers, content and dates. (published 2026-10-16, last reviewed 2026-10-16) - https://northend.law/guides/how-to-run-a-dpia - How to run a DPIA in practice, with a checklist: A data protection impact assessment step by step: screening, describing the processing, necessity, risks, measures, consultation, decision and review. (published 2026-10-16, last reviewed 2026-10-16) - Also covered: https://northend.law/guides/ab-testing#legitimate-interest-assessments - "Legitimate interest assessments" in A/B Testing ### EU AI Act and US AI rules (https://northend.law/guides/eu-ai-act-and-us-ai-rules) The EU AI Act and US state rules on AI: who is in scope, the duties of providers and deployers, and transparency for chatbots and assistants. - https://northend.law/guides/eu-ai-act-timeline - EU AI Act timeline after the omnibus: what applies when: The EU AI Act dates in force after the 2026 omnibus: prohibitions, AI literacy, general-purpose models, transparency, and the new high-risk dates. (published 2026-10-16, last reviewed 2026-10-16) - https://northend.law/guides/ai-compliance - AI compliance: An introduction to AI compliance duties under the EU AI Act, the Colorado AI Act and similar rules: transparency, documentation and human oversight. (published 2026-01-23, last reviewed 2026-01-23) - https://northend.law/guides/customer-facing-chatbots - Customer-facing Chatbots: What a business deploying an AI chatbot should consider: the FTC's focus, telling users they are talking to AI, deceptive practices, and conversation data. (published 2026-01-28, last reviewed 2026-01-28) - https://northend.law/guides/ai-assistants - AI Assistants: Legal risks of enterprise AI assistants and copilots: confidential data, acceptable use policies, liability for outputs, and documentation and audit trails. (published 2026-01-28, last reviewed 2026-01-28) - https://northend.law/guides/brain-computer-interfaces - Brain Computer Interfaces: How new laws protect neural data from brain-computer interfaces: dedicated statutes, heightened consent, FDA device considerations, and ethical questions. (published 2026-01-28, last reviewed 2026-01-28) ### International data transfers (https://northend.law/guides/international-data-transfers) Moving personal data out of the EU and the UK: the Data Privacy Framework, standard contractual clauses, transfer impact assessments and the UK addendum. - https://northend.law/guides/eu-transfers-after-dpf - Transferring personal data out of the EU after the Data Privacy Framework: SCCs, TIAs and the US adequacy decision: Sending personal data from the EU to the US: when the Data Privacy Framework covers it, when you need SCCs and a transfer impact assessment, and transfers to AI providers. (published 2026-10-16, last reviewed 2026-10-16) - https://northend.law/guides/uk-and-swiss-transfers - UK and Swiss transfers: the IDTA, the UK Addendum and the Swiss list: How to transfer personal data out of the UK and Switzerland: the IDTA, the UK Addendum, the UK Extension for US companies, the new UK test, and the Swiss list. (published 2026-10-16, last reviewed 2026-10-16) - Also covered: https://northend.law/guides/digital-analytics#cross-border-data-transfer-compliance - "Cross-border data transfer compliance" in Digital Analytics ### Privacy-enhancing technologies (https://northend.law/guides/privacy-enhancing-technologies) What data clean rooms, identity solutions, customer data platforms and other privacy-enhancing technologies do in law. - https://northend.law/guides/data-clean-rooms - Data Clean Rooms: What data clean rooms can and cannot do for privacy, the FTC's warnings about marketing claims, GDPR and CCPA questions, and the contract terms to use. (published 2026-01-28, last reviewed 2026-01-28) - https://northend.law/guides/identity-solutions - Identity Solutions (AdTech): Privacy questions raised by adtech identity solutions such as UID2, ID5 and RampID: consent to create identifiers, and tracking across devices and sites. (published 2026-01-28, last reviewed 2026-01-28) - https://northend.law/guides/customer-data-platforms - Customer Data Platforms: How a customer data platform affects privacy compliance: the lawful basis for unifying data, consent management, and due diligence on the CDP vendor. (published 2026-01-28, last reviewed 2026-01-28) - Publication (2025-02-01): https://northend.law/publications/avoiding-the-scope-of-the-eprivacy-directive-in-advertising - Avoiding the scope of the ePrivacy Directive in advertising - Publication (2024-06-06): https://northend.law/publications/pepr24-takeaways - Some takeaways from PEPR'24 (USENIX Conference on Privacy Engineering Practice and Respect) ### AI-native law firms: a reference (October 2026) - https://northend.law/guides/ai-native-law-firms-reference - AI-native law firms: a reference (October 2026): An unranked comparison of firms that call themselves AI-native: structure, admissions, main work, published prices and AI, from each firm's own site. (published 2026-10-16, last reviewed 2026-10-29) ### Copyright (https://northend.law/copyright) - https://northend.law/copyright/trademarks - Trademarks: How a startup protects its brand: searching before choosing a name, filing early in key markets, choosing a strong mark, and watching for infringement. (published 2026-01-23, last reviewed 2026-01-23) - https://northend.law/copyright/copyright-basics - Copyright: Who owns the code and content a company creates: work-for-hire terms with contractors, open source licence duties, and when registration matters. (published 2026-01-23, last reviewed 2026-01-23) ## Pricing Philosophy Transparent, predictable pricing. Two tracks: 1. **Subscription-based services** (recurring monthly/quarterly retainers) 2. **Project-based legal advice** (ad-hoc engagements; attorney-priced lanes state the exact flat fee in the engagement letter before signing) Plans are scoped to complement an existing GC or law firm rather than replace one. No billable hours; no surprise invoices. Machine time is accounted as the firm's cost of production, never billed as hours. ## Key Differentiators - **Software included free**: Clients receive the platform (DPO Central, AI Sentinel, vendor.watch, and Dealroom), built by our technology company, at no extra cost, so they pay only for our lawyers' judgment. - **AI-native operations with tiered trust**: privileged work runs only on self-hosted models on firm hardware; confidential work runs in a secure private cloud on open-weight models; commercial frontier models handle only abstract, carefully anonymized research. For engagement work, customer-specific or confidential information is not sent to proprietary cloud LLMs or commercial "wrappers." The contract upload at /contract-review follows the same rule: the file is stored with the site's hosting provider and read by a lawyer, never by an AI model. - **Cross-jurisdictional**: The founding attorney has been admitted in California (US), Spain (Abogado, Ilustre Colegio de la Abogacía de Madrid, EU), and England & Wales (Solicitor); cross-border practice across US/UK/EU. - **Complementary, not competitive**: Designed to plug into a client's existing legal stack alongside their current go-to firm. ## Target Keywords - AI-native law firm - Outside counsel subscription - AI governance lawyer - EU AI Act compliance - NIST AI RMF - ISO 42001 - Agentic contracts - MCP contracts - Agent-to-agent transactions - A2A dispute resolution - Token structuring lawyer - DeFi compliance - Smart contract law - GDPR lawyer - Cross-border privacy ## Leadership **Sergio Maldonado Elvira** (LL.M., J.D., FIP, CIPP/US, CIPP/E, CIPT) - Founder & Controlling Shareholder - Admitted in California (Attorney, State Bar of California) - Admitted in Madrid, Spain (Abogado, Ilustre Colegio de la Abogacía de Madrid, since 1998) - Solicitor, Law Society of England & Wales (2003-2007) - IAPP certifications: FIP, CIPP/US, CIPP/E, CIPT - Degrees: LL.M., J.D. - Profile: https://northend.law/leadership - Member of the European Data Protection Board's "Pool of Experts" ## Contact - Website: https://northend.law - Start an engagement: https://northend.law/start - Contact & booking: https://northend.law/contact - Phone: +1 737 336 4630 - Office: 447 Sutter St. Suite 506 - 1349, San Francisco CA 94108, United States - Law Corporation number 31414 - Approved by The State Bar of California ## Site Structure - / - Homepage - /ai - How the firm uses AI: self-hosted stack, tiered trust, citation verification - /submit-a-matter - Submit a matter in one screen, no account (privacy, AI governance, the AI-native work around them, and commercial contracts in MarTech, AdTech and SaaS) - /start - Start an engagement (six lanes, conflicts check, e-signed engagement letter) - /pricing - Subscription & project-based pricing - /dpoaas - Privacy Counsel detail page (quarterly tiers, slider) - /platform - The Platform (free software included with any engagement) - /skills - Skills & Knowledge Bases (attorney-built AI review Skills + provenance KBs; free gated samples) - /ai-governance-plan - Fractional AI Governance Lead detail page - /privacy - Privacy & Data Protection practice area - /mobile-apps - Mobile App Privacy and Compliance (policies, terms, store declarations, SDKs; health data and minors) - /copyright - Copyright & Licensing practice area - /ai-governance - Global AI Governance practice area - /agentic-contracts - Agentic Contractual Frameworks practice area - /ai-crypto - AI-Crypto Interplay practice area - /crypto - Crypto and digital assets practice area (wallet licensing, stablecoins, MiCA, travel rule, US money transmission, UK regime) - /contract-review - Contract review by an attorney (upload a draft) - /leadership - About the founder - /publications - Selected articles (privacy, data protection, AI governance, legal practice), republished from Medium, LinkedIn and print - /podcast - Podcast - /guides - Guides: topic hubs on privacy, AI governance and technology contracts (listed above) - /deep-dives - Crypto and digital-asset analyses (wallet custody, US money transmission, UK regime, stablecoins, DAOs) - /prompts - Curated AI prompt library (for clients' own AI use) - /scans - Free website exposure scan: every ad-tech vendor, pixel, cookie and tracking ID on a company's public site, screened against U.S. privacy litigation records, delivered as a counsel-reviewed report - /inscope - InScope regulatory radar: which privacy, data protection and AI governance laws apply to a company (29 instruments, US/EU/UK and beyond), compressed into shared actions; embedded interactive sample on the page, first report free via the request form - /contact - Contact and booking ## Customer-facing app subdomains (software included free with any engagement) - https://dpocentral.todo.law - DPO Central (privacy management) - https://aisentinel.todo.law - AI Sentinel (AI governance) - https://vendor.watch - vendor.watch (third-party/vendor risk) - https://dealroom.todo.law - Dealroom (contract negotiation) ## For AI agents: MCP server - Endpoint: https://northend.law/api/mcp (MCP, streamable HTTP, no sign-in for the free tools). Agent card: https://northend.law/.well-known/agent-card.json. Page: https://northend.law/agents - Service: Data processing agreement, $350 flat fee (United States dollars), paid in full at checkout; No fee for current Privacy Counsel subscribers. Governing law: California by default, or Spain on request. Companies established in the United States, the European Union or the United Kingdom. Countries under sanctions are refused. - Service: Business associate agreement (HIPAA), $350 flat fee (United States dollars), paid in full at checkout; No fee for current Privacy Counsel subscribers. For a vendor whose services are designed not to receive protected health information. Governing law: California. Companies established in the United States. - Free tools: list_services, get_requirements, get_engagement_terms, request_consultation (get_dpa_requirements remains as the DPA's alias). Matter tools: open_matter, get_matter_status, get_payment_link, submit_facts, get_deliverable (submit_dpa_facts remains as an alias). - Target, not a guarantee: the agreement is released within 4 business hours of complete facts, after a lawyer's review. Business hours are 9:00 to 17:00 Pacific time, Monday to Friday, except United States federal holidays. - Responsible lawyer: Sergio Maldonado Elvira, Founder & Principal Attorney, North End Law Corporation, 447 Sutter St. Suite 506 - 1349, San Francisco, California 94108, United States. - Using these tools does not create a lawyer-client relationship. The firm accepts the engagement when its automatic conflicts check is clear and payment is received (for a subscriber, when the engagement is accepted and the check is clear). If a possible conflict is found, nothing is charged, or the payment is refunded in full. Do not send confidential facts before then. ## Data processing agreement AI agents can ask North End Law for two legal services on behalf of a company, each for a flat fee of $350: a data processing agreement and a business associate agreement. No fee for current Privacy Counsel subscribers. ### Price $350 flat fee per agreement, United States dollars, paid in full at checkout. No fee for current Privacy Counsel subscribers. ### Includes - One data processing agreement between your company and the named counterparty, under California law, or Spanish law on request. - Review by a lawyer of the firm before release. - One round of revisions within 30 days of release. ### Excludes - Negotiation with the counterparty. - Review of a draft proposed by the counterparty. - Advice beyond the agreement itself. - Disputes, regulatory filings or investigations. ### Steps 1. open_matter with the service, the governing law and the client's role only, and no names. It returns a private link: give it to the user, who enters the parties' details there. The firm runs an automatic conflicts check. 2. The person who signs receives the engagement text by e-mail and accepts it with a one-time code. 3. get_payment_link: the fee is charged in full at checkout. Once payment is received and the conflicts check is clear, the engagement has begun. 4. submit_dpa_facts with the facts listed by get_dpa_requirements, without names: "the Client" and "the Customer" (or "the Vendor"). 5. A lawyer reviews the agreement and releases it. get_deliverable gives a copy with placeholders; the final agreement, with the parties' details, goes to the person who signs, by e-mail and on the private page. ## Business associate agreement Companies established in the United States. ### Price $350 flat fee per agreement, United States dollars, paid in full at checkout. No fee for current Privacy Counsel subscribers. ### Includes - One business associate agreement under HIPAA (45 C.F.R. 164.504(e)) between your company and the named counterparty, under California law. - For a vendor whose services are designed not to receive protected health information: the agreement applies only if the vendor ever does receive it. - Review by a lawyer of the firm before release. - One round of revisions within 30 days of release. ### Excludes - Negotiation with the counterparty. - Review of a draft proposed by the counterparty. - Advice beyond the agreement itself. - Disputes, regulatory filings or investigations. - A HIPAA risk analysis, security policies, state health-privacy laws or 42 C.F.R. Part 2 records. - A vendor whose services are designed to handle protected health information: use request_consultation. ### Steps 1. open_matter with tool "baa", the governing law, the client's role and receivesPhi false only, and no names. It returns a private link: give it to the user, who enters the parties' details there. The firm runs an automatic conflicts check. 2. The person who signs receives the engagement text by e-mail and accepts it with a one-time code. 3. get_payment_link: the fee is charged in full at checkout. Once payment is received and the conflicts check is clear, the engagement has begun. 4. submit_facts with the facts listed by get_requirements for "baa", without names: "the Client" and "the Customer". The facts must not contain protected health information. 5. A lawyer reviews the agreement and releases it. get_deliverable gives a copy with placeholders; the final agreement, with the parties' details, goes to the person who signs, by e-mail and on the private page. ### Safeguards - The firm never gives your assistant the parties' names or addresses: you enter them on a private page. - The firm accepts the engagement when its automatic conflicts check is clear and payment is received (for a subscriber, when the engagement is accepted and the check is clear). If a possible conflict is found, nothing is charged, or the payment is refunded in full. - The person who signs accepts the engagement by a one-time code sent to their work e-mail, and can stop a request that was not theirs. - The client's facts are encrypted at rest, are never sent to any artificial intelligence model, and are deleted from this site 30 days after release. - A lawyer reviews every agreement before release. ### Tools - list_services: What North End Law offers through this server, at what fixed price, under which laws, and who is responsible. Responsible lawyer: Sergio Maldonado Elvira, Founder & Principal Attorney, North End Law Corporation, 447 Sutter St. Suite 506 - 1349, San Francisco, California 94108, United States. - get_requirements: The facts a service needs, as a JSON Schema: service "dpa" (data processing agreement) or "baa" (business associate agreement). For a DPA, optionally narrowed to one governing law and the client's role. Do not send the facts until the engagement has begun. - get_dpa_requirements: The facts the data processing agreement needs, as a JSON Schema. Optionally narrowed to one governing law and the client's role. Do not send the facts until the engagement has begun. - get_engagement_terms: The full limited-scope engagement text for a service, with its version and SHA-256 hash, and whether it is open for acceptance. - request_consultation: For anything outside the listed services: gives the booking link and lets the firm know a person would like to talk. Send a one-line topic only, no confidential facts. - open_matter: Starts a request for a data processing agreement ($350) or, with tool "baa", a business associate agreement ($350). Send only the service, the governing law (CALIFORNIA by default, or SPAIN; CALIFORNIA only for a BAA), the client's role (PROCESSOR by default, or CONTROLLER; BUSINESS_ASSOCIATE for a BAA, with receivesPhi false) and, if useful, a short label that names no one. Do not ask the user for, and do not send, any name, e-mail address or postal address: the firm never needs them from you. Write "the Client" for the user's company and "the Customer" (or "the Vendor", when the client is the controller) for the other party. Returns a matter id, a matter token (shown once: keep it) and a private link. Give the private link to the user and ask them to open it in their browser within 24 hours. There the user enters their company's details, their own and the other party's, and the firm runs its conflicts check. You never see those details. The firm accepts the engagement when its automatic conflicts check is clear and payment is received (for a subscriber, when the engagement is accepted and the check is clear). If a possible conflict is found, nothing is charged, or the payment is refunded in full. A vendor whose services handle protected health information should use request_consultation. - get_matter_status: Where the matter stands, the next human step, the target in business hours, and any question from the lawyer. - get_payment_link: The firm's own card payment page for $350, available once the named person has accepted the engagement. The fee is charged in full at checkout and refunded in full if the firm declines the matter or does not deliver. The link is also e-mailed to the named person. - submit_facts: The fact package for the matter's service, in the "dealroom.solo-intake/1" shape described by get_requirements. Accepted only once the engagement has begun, and again when the lawyer asks for changes. Leave out counterpartyLegalName: the firm adds the other party from the private page. Do not ask the user for, and do not send, any name, e-mail address or postal address: the firm never needs them from you. Write "the Client" for the user's company and "the Customer" (or "the Vendor", when the client is the controller) for the other party. A fact package that contains a name, an e-mail address or a postal address the firm holds for the matter is refused. For a business associate agreement, the facts must not contain protected health information: they describe the services, not any individual. - submit_dpa_facts: The fact package for the agreement, in the "dealroom.solo-intake/1" shape described by get_dpa_requirements. Accepted only once the engagement has begun, and again when the lawyer asks for changes. Leave out counterpartyLegalName: the firm adds the other party from the private page. Do not ask the user for, and do not send, any name, e-mail address or postal address: the firm never needs them from you. Write "the Client" for the user's company and "the Customer" (or "the Vendor", when the client is the controller) for the other party. A fact package that contains a name, an e-mail address or a postal address the firm holds for the matter is refused. - get_deliverable: Once a lawyer has released it: the agreement in Markdown with placeholders ([Client], [Customer] or [Vendor], and blank addresses and signatures), so you can explain it to the user. The final agreement with the parties' details, in PDF and Word, is only on the private page and in the e-mail to the person who signs. ### Discovery - https://northend.law/.well-known/mcp.json - https://northend.law/.well-known/mcp/server-card.json - https://northend.law/.well-known/agent-card.json - https://northend.law/agents Responsible lawyer: Sergio Maldonado Elvira, Founder & Principal Attorney, North End Law Corporation, 447 Sutter St. Suite 506 - 1349, San Francisco, California 94108, United States. Using these tools does not create a lawyer-client relationship. The firm accepts the engagement when its automatic conflicts check is clear and payment is received (for a subscriber, when the engagement is accepted and the check is clear). If a possible conflict is found, nothing is charged, or the payment is refunded in full. Do not send confidential facts before then. ## Guides: full text ### Received a CIPA demand letter about your website? What to do first URL: https://northend.law/guides/cipa-demand-letter Section: Website and app legal documents By Wenceslao M. Spiegel. Published 2026-10-16; last reviewed 2026-10-16. General information, not legal advice. A CIPA demand letter claims that a tool on your website (a pixel, a chat widget, a session replay script or a tracker) let a third party intercept or record visitors' communications without their consent, in breach of the California Invasion of Privacy Act, and asks for a payment to settle before a lawsuit or an arbitration is filed. Do not ignore it and do not answer it yourself: preserve the evidence, find out what your site actually ran on the dates claimed, check your vendor contracts and insurance, and have counsel answer. ## What does the letter claim? Most letters rely on one or both of two sections of the California Penal Code: - Section 631(a), the wiretapping provision. It reaches anyone who, without the consent of all parties, reads or attempts to read or to learn the contents of a communication, and anyone who aids or permits that. Letters argue that a vendor's script on your site learns what visitors type or click, and that you aided it. - Section 638.51, the pen register and trap and trace provision. It prohibits installing or using a pen register or a trap and trace device without first obtaining a court order, except as specified. Letters argue that a tracker that collects IP addresses or device information is such a device. The letter usually points to Section 637.2, which gives a person injured by a violation a civil claim. Relief includes the greater of $5,000 per violation or three times actual damages, and an injunction, and the plaintiff need not have suffered actual damages. That figure, multiplied across visitors, is what gives the letter its weight. Sources: Penal Code § 631 (https://leginfo.legislature.ca.gov/faces/codes_displaySection.xhtml?lawCode=PEN§ionNum=631); Penal Code § 638.51 (https://leginfo.legislature.ca.gov/faces/codes_displaySection.xhtml?lawCode=PEN§ionNum=638.51); Penal Code § 637.2 (https://leginfo.legislature.ca.gov/faces/codes_displaySection.xhtml?lawCode=PEN§ionNum=637.2); Assembly Committee on Privacy and Consumer Protection, analysis of SB 690 (https://apcp.assembly.ca.gov/system/files/2026-06/sb-690-caballero-apcp-analysis.pdf). ## What does SB 690 change? The Governor signed SB 690 on 30 September 2026. In his words, it "eliminates the private right of action under the California Invasion of Privacy Act (CIPA) for violations of the pen register and trap-and-trace statute arising from conduct occurring on an internet website, online application, or mobile application." Under the California Constitution, a statute enacted at a regular session goes into effect on 1 January next following a 90-day period from its enactment, unless one of the Constitution's exceptions (such as an urgency statute) applies (article IV, section 8(c)). For a bill signed on 30 September 2026, that is 1 January 2027. In place of private claims, the Attorney General may enforce the pen register provision for this conduct. The bill also reaches some cases already filed: its author states that it "applies the change to qualifying claims filed during the previous two years", and the text proposed in the Assembly committee applies it to "any pending claim in an action commenced within two years before the operative date". Counsel should read the enacted text against the filing date of the case. What SB 690 does not change: the private claim under Section 631 (wiretapping) and the other sections of CIPA remain. A letter that relies on Section 631 is not answered by SB 690, and many letters plead both. The Governor's message also asked the Legislature to look at the other CIPA provisions next year. Sources: Governor of California, legislative update of 30 September 2026 (https://www.gov.ca.gov/2026/09/30/governor-newsom-issues-legislative-update-9-30-2026/); signing message for SB 690 (https://www.gov.ca.gov/wp-content/uploads/2026/09/SIGN-msg-SB-690.pdf); California Constitution, article IV (Assembly Clerk) (https://clerk.assembly.ca.gov/sites/clerk.assembly.ca.gov/files/ARTICLE_IV_Session_2017-18.pdf); SB 690, California Legislative Information (https://leginfo.legislature.ca.gov/faces/billNavClient.xhtml?bill_id=202520260SB690); Senator Caballero, release of 31 August 2026 (https://sd14.senate.ca.gov/node/1044); Assembly Committee on Privacy and Consumer Protection, analysis of SB 690, comment 7 (https://apcp.assembly.ca.gov/system/files/2026-06/sb-690-caballero-apcp-analysis.pdf). ## What should you do first? 1. Note the deadline, but do not reply yourself. Letters set short deadlines. A quick reply that explains or apologises can be used as an admission. Send the letter to counsel and to your insurer (step 5). 2. Preserve the evidence. Issue a written hold to the people who manage the website, the tag manager and the consent tool. Keep the site's configuration, the tag manager's version history and the consent banner settings as they stood on the dates the letter names. Do not delete scripts or logs before they are preserved; removing a tool going forward is a separate decision for counsel. 3. Find out what the site actually ran. List every script, pixel and cookie that loaded on the pages named in the letter, and whether each one fired before or after the visitor's choice in the banner. The letter's description of your site is often generic. This firm offers a free scan of a public website for this purpose (website scan (/scans)); because the scan is offered by this firm, we mention it with that disclosure. 4. Check the vendors. For each tool named, find the contract, the data processing terms and any indemnity. Note whether the vendor may use the data for its own purposes or only for yours. Whether a vendor that uses the data only for the website owner is a third party to the communication is disputed; one federal court treated such a vendor as an extension of the website owner rather than a third party (Graham v. Noom, Inc., No. 3:20-cv-06903-LB (N.D. Cal. Apr. 8, 2021) (https://www.govinfo.gov/content/pkg/USCOURTS-cand-3_20-cv-06903/pdf/USCOURTS-cand-3_20-cv-06903-0.pdf)). 5. Notify your insurers. Cyber, media liability and directors' and officers' policies may respond, and each has its own notice conditions. Some policies exclude claims based on wiretapping or privacy statutes; read the exclusions with counsel before relying on cover. 6. Check your own website terms. If your terms of use contain an arbitration clause and a class action waiver, whether they bind the visitor depends on how the visitor accepted them (a click, or only a link in the footer). See our guide on terms of service (/guides/terms-of-service). 7. Have counsel answer. Counsel can test the letter's facts against the evidence from steps 2 to 4, assess which claims survive SB 690, and decide whether to answer, negotiate or wait. ## Pixels, chat widgets and session replay: which theory reaches which tool? Tool | Theory usually pleaded | What to check Advertising pixels (social media and search platforms) | Section 631 (the platform learns which pages a visitor views or what a visitor searches); Section 638.51 (the pixel collects IP address and device data) | Whether the pixel fired before consent; what data it sent; whether the platform may use the data for its own advertising Chat widgets | Section 631 (a third party reads or records the chat as it is typed) | Who hosts the chat; whether the vendor may use transcripts for its own products; the notice shown before the chat starts Session replay | Section 631 (keystrokes and clicks recorded as they happen) | Masking of form fields; whether recording starts before consent; the vendor's own use of recordings Analytics and beacon scripts | Section 638.51 (addressing and device information) | SB 690 removes the private claim for website and app conduct; check whether the case falls within its two-year look-back ## What a cookie banner does, and does not, do A banner helps only if the tools actually wait for the visitor's choice. A banner that shows "reject" but lets the pixel fire on page load supports the letter's claim rather than answering it. Consent also has to come first: in an unpublished decision, the Ninth Circuit held that Section 631(a) requires the prior consent of all parties, so agreement given later in a visit does not cure what was recorded before it (Javier v. Assurance IQ, LLC, No. 21-16351 (9th Cir. May 31, 2022)). Unpublished decisions of that court are not precedent except as its Rule 36-3 provides. Source: Javier v. Assurance IQ, LLC, memorandum of 31 May 2022 (https://cdn.ca9.uscourts.gov/datastore/memoranda/2022/05/31/21-16351.pdf). ## Checklist - Letter sent to counsel and to the insurer within the policy's notice period - Written hold issued; site configuration, tag manager history and banner settings preserved - List of every script and pixel on the named pages, with firing order against consent - Vendor contracts, data processing terms and indemnities collected - Each claim mapped to Section 631 or Section 638.51, and to SB 690 - Website terms checked: arbitration clause, class waiver, how visitors accepted them - No reply sent by the company itself ## Questions people ask Should we just pay? Not before counsel has compared the letter with what your site actually ran. A letter may describe tools the site never used, or raise claims that SB 690 removes. A payment made without a release drafted by counsel may not end the matter. How long do we have? The deadline in the letter is the sender's, not the law's. It still matters for the negotiation, and your insurance policy may set its own notice period, so act within days. Does SB 690 end our exposure? Only for private claims under the pen register provision based on website or app conduct. Claims under Section 631 continue. Will our insurance cover it? It may. Cyber and media liability policies sometimes respond to privacy claims, and some exclude wiretapping statutes. Notify early and let counsel read the exclusions. Is removing the pixel an admission? Changing the site going forward is not in itself an admission, but preserve the earlier configuration first and let counsel decide the timing. ## Update log - 16 October 2026: first published. Sources checked the same day; SB 690's look-back for filed cases and a court decision on vendors added. ### Privacy policies URL: https://northend.law/guides/privacy-policies Section: Website and app legal documents By Wenceslao M. Spiegel. Published 2026-01-23; last reviewed 2026-01-23. General information, not legal advice. Your privacy policy is a legally binding document and a trust signal to users. It must accurately describe your practices while remaining accessible to non-lawyers. Key points: - Cover all data collection touchpoints accurately - Explain user rights under GDPR and CCPA - Keep policies updated as practices evolve - Make policies accessible and easy to understand #### Describe actual practices accurately Your policy must match reality. If you collect location data, say so. If you share data with advertisers, disclose it. Regulators impose penalties for deceptive practices, saying one thing while doing another. Conduct a data mapping exercise before drafting. #### Cover all required disclosures Different laws require different disclosures. GDPR needs: legal basis, retention periods, data subject rights, transfer mechanisms, DPO contact. CCPA needs: categories of data, purposes, sale/sharing practices, consumer rights. Layer your policy to address multiple jurisdictions. #### Make it readable Legal doesn't mean inaccessible. Use clear headings, plain language, and practical examples. Consider a layered approach: a short-form summary linking to detailed sections. Users who can understand your policy are more likely to trust your practices. #### Keep it current Policies become stale. When you add new features, integrations, or marketing tools, update the policy. When laws change, update the policy. Maintain version history and notify users of material changes. An outdated policy creates compliance risk. ### Terms of service URL: https://northend.law/guides/terms-of-service Section: Website and app legal documents By Wenceslao M. Spiegel. Published 2026-01-23; last reviewed 2026-01-23. General information, not legal advice. Your terms of service establish the contractual relationship with users. They protect your interests, set expectations, and provide mechanisms for resolving disputes. Key points: - Define acceptable use and prohibited conduct - Include limitation of liability provisions - Reserve rights to modify terms with notice - Add dispute resolution and governing law clauses #### Define acceptable use clearly Specify what users can and cannot do. Prohibited conduct should cover: illegal activity, abuse, reverse engineering, and anything that harms your service or other users. Be specific enough to enforce but broad enough to cover evolving threats. #### Protect yourself with liability limits Include limitation of liability clauses capping your exposure, typically to fees paid. Disclaim warranties to the extent permitted (varies by jurisdiction). Include indemnification from users for their content and conduct. These provisions are essential risk management. #### Reserve rights to change and terminate Your product will evolve. Reserve the right to modify terms with reasonable notice. Include the right to suspend or terminate accounts for violations. Specify what happens to user data upon termination. #### Choose your dispute resolution mechanism Courts are expensive and unpredictable. Many companies require arbitration (faster, cheaper, more predictable) and class action waivers (prevents bet-the-company litigation). Include governing law and venue provisions. Consider whether to allow small claims court as an alternative. ### Cookies, pixels & trackers URL: https://northend.law/guides/cookies-pixels-trackers Section: Website and app legal documents By Wenceslao M. Spiegel. Published 2026-01-23; last reviewed 2026-01-23. General information, not legal advice. Cookies, pixels, and trackers are under intense regulatory scrutiny. Getting consent right is both a legal obligation and a user experience consideration. Key points: - Implement proper consent management platform - Categorize trackers by purpose and necessity - Block non-essential trackers before consent - Document tracker vendors and data flows #### Implement a proper consent platform Cookie banners with only an 'Accept' button don't cut it. You need a full Consent Management Platform (CMP) offering: granular category choices, easy rejection, preference persistence, and consent logging. Tools like OneTrust, Cookiebot, or open source options work. #### Categorize trackers correctly Not all cookies are equal. Strictly necessary cookies (authentication, shopping cart) don't need consent. Everything else does: analytics (performance), advertising (targeting), social media. Categorize accurately, miscategorization invites enforcement. #### Block before consent Non-essential trackers must not fire until users consent. This requires technical implementation: tag management with consent mode, scripts that wait for consent signals. Soft consent (assumed from scrolling) is not valid under GDPR. #### Document your tracker ecosystem Maintain a tracker inventory: what it is, who provides it, what data it collects, where that data goes. This feeds your privacy policy disclosures and audit responses. Review regularly as your marketing stack evolves. ### Session Replay Tools URL: https://northend.law/guides/session-replay-tools Section: Website and app legal documents By Wenceslao M. Spiegel. Published 2026-01-28; last reviewed 2026-01-28. General information, not legal advice. Session replay tools that record user interactions face significant legal scrutiny, including wiretapping claims and privacy regulation enforcement. Proper implementation requires careful consent management and data masking. Key points: - Obtain proper consent before recording - Mask sensitive data automatically - Address wiretapping law concerns - Manage litigation and enforcement risks #### Wiretapping litigation risk Session replay tools have faced class action lawsuits under wiretapping statutes, particularly California's CIPA. Plaintiffs argue that recording keystrokes, mouse movements, and form inputs without consent constitutes unlawful interception. Several cases have survived motions to dismiss. While outcomes vary, the litigation risk is real and can result in statutory damages. #### Consent requirements across jurisdictions GDPR requires consent for non-essential session recording, it's not strictly necessary for website functionality. US state privacy laws may require disclosure and opt-out. Wiretapping laws may require consent from 'all parties' to the communication. Implement clear, affirmative consent before session recording begins, not just privacy policy disclosure. #### Data masking and sensitive information Session replays can capture sensitive data users enter in forms: passwords, payment information, personal details. Configure aggressive masking: automatically redact password fields, mask credit card numbers and SSNs, exclude healthcare information forms, and consider masking all text input by default. Audit recordings regularly to catch masking failures. #### Vendor selection and configuration Session replay vendor choice affects your compliance posture. Evaluate: default privacy settings (opt-in masking vs. opt-out), data storage locations and retention periods, sub-processor lists and data sharing, consent integration capabilities, and contractual commitments. Many breaches stem from misconfigured defaults, review settings carefully. ### Conversion APIs URL: https://northend.law/guides/conversion-apis Section: Website and app legal documents By Wenceslao M. Spiegel. Published 2026-01-28; last reviewed 2026-01-28. General information, not legal advice. Conversion APIs (CAPIs) send event data server-to-server rather than through browser pixels, improving data quality while raising distinct privacy compliance questions. Key points: - Ensure consent flows to server events - Maintain data minimization principles - Document data flows and retention - Navigate platform-specific requirements #### CAPI privacy fundamentals Server-side tracking doesn't bypass privacy requirements, it changes where they're implemented. CAPIs still process personal data (email hashes, IP addresses, transaction details) and require lawful basis. The key difference: browsers don't see the tracking, but legal obligations remain identical. Don't confuse technical invisibility with privacy compliance. #### Consent signal propagation When users decline tracking consent, that preference must flow through to your CAPI implementation. Technical challenges include: real-time consent signal sync to servers, preventing event transmission before consent confirmation, handling consent revocation for pending events, and maintaining consent audit trails server-side. Test thoroughly, consent gaps are a common CAPI pitfall. #### Data minimization in CAPI events CAPIs can transmit extensive data including PII like email addresses. Apply data minimization: only send data actually needed for conversion attribution, hash identifiers appropriately, avoid transmitting sensitive transaction details unless necessary, and configure appropriate data retention on receiving platforms. More data isn't better if it's not justified. #### Platform-specific requirements Each platform's CAPI has distinct requirements: Meta requires specific data processing terms, Google has ads data processing terms, and various platforms impose technical requirements for data formatting and transmission. Review and execute required legal agreements before implementation. Consider unified CAPI management solutions that handle multi-platform compliance. ### Digital Analytics URL: https://northend.law/guides/digital-analytics Section: Website and app legal documents By Wenceslao M. Spiegel. Published 2026-01-28; last reviewed 2026-01-28. General information, not legal advice. Digital analytics, measuring website and app performance, is foundational to business operations but faces ongoing regulatory challenges. Compliant analytics requires careful tool selection, configuration, and consent management. Key points: - Configure analytics for GDPR compliance - Evaluate privacy-friendly alternatives - Manage consent for tracking cookies - Navigate data transfer requirements #### Google Analytics compliance challenges Google Analytics has faced regulatory action across Europe due to US data transfers. Following Schrems II, several EU DPAs found GA4 implementations non-compliant. While Privacy Shield 2.0 (EU-US Data Privacy Framework) provides a new legal mechanism, its durability is uncertain. Evaluate whether GA-specific legal risks are acceptable for your organization. #### Privacy-friendly analytics alternatives Alternatives to traditional analytics include: first-party analytics (Plausible, Fathom, Matomo) that don't share data with third parties, aggregate measurement solutions that minimize personal data, server-side log analysis avoiding client-side tracking, and privacy-preserving measurement APIs (Attribution Reporting API). Evaluate tradeoffs between data richness and privacy risk. #### Consent implementation for analytics Under GDPR, analytics cookies generally require consent, they're not strictly necessary for website functionality. Implement proper consent flows: block analytics scripts until consent obtained, respect withdrawal by stopping data collection, maintain consent records, and audit consent rates to understand data coverage. Cookie-less analytics approaches may reduce consent friction. #### Cross-border data transfer compliance Analytics data sent to US-based providers raises transfer mechanism questions. Options include: relying on EU-US Data Privacy Framework certification, implementing Standard Contractual Clauses with transfer impact assessments, using EU-based analytics providers, or configuring existing tools to store data regionally. Document your legal basis for any cross-border transfers. ### A/B Testing URL: https://northend.law/guides/ab-testing Section: Website and app legal documents By Wenceslao M. Spiegel. Published 2026-01-28; last reviewed 2026-01-28. General information, not legal advice. A/B testing occupies a gray area under privacy regulations, sometimes essential for website functionality, sometimes requiring consent. Proper classification and documentation are key to compliant experimentation. Key points: - Classify testing as essential vs. optional - Document legitimate interest assessments - Handle consent for personalization tests - Avoid dark pattern implications #### A four-step check before you run a test 1. Does the test store or read anything on the user's device (a cookie, local storage, an SDK)? For users in the EU or the UK, that needs consent unless it is strictly necessary for the service the user asked for. 2. Does it use personal data? Most product tests can rely on legitimate interest, recorded in a short assessment. Where the test depends on tracking that already needs consent, the test needs it too. 3. Does it personalize by profile? Tell users in your privacy notice, and keep sensitive data (health, religion, sexuality and similar) out of the test. 4. Does a vendor run it? Put a DPA in place. In California, the vendor must be bound as a service provider, or the data flow may count as a "sale" or "sharing" that users can opt out of. #### Essential vs. non-essential testing Not all A/B tests require consent. Tests for legitimate website functionality, load time optimization, error reduction, accessibility improvements, may qualify as essential under GDPR's legitimate interest basis. Tests for conversion optimization, personalization, or marketing effectiveness are typically non-essential requiring consent. Document the purpose and basis for each testing program. #### Legitimate interest assessments If relying on legitimate interest for testing, conduct proper balancing tests. Document: the purpose of testing and business benefit, impact on individuals (minimal for UI tests, greater for personalization), user expectations given context, safeguards implemented (data minimization, pseudonymization), and why consent is impractical. Maintain these assessments as records. #### Personalization testing considerations Tests involving personalization based on user profiles require more careful analysis. Creating user segments, testing content tailored to inferred preferences, or experimenting with pricing may require consent under GDPR's automated decision-making provisions or CCPA's profiling disclosures. Consider whether your testing involves 'significant effects' on individuals. #### Avoiding dark pattern claims A/B tests that optimize for conversions can inadvertently create dark patterns, interface designs that manipulate users against their interests. Regulators increasingly scrutinize such practices. Tests should improve user experience, not exploit cognitive biases. Document ethical guidelines for testing and review concerning test variations. ### Health Apps (Mobile) URL: https://northend.law/guides/health-apps Section: Website and app legal documents By Wenceslao M. Spiegel. Published 2026-01-28; last reviewed 2026-01-28. General information, not legal advice. Mobile health apps operate in a complex regulatory landscape where HIPAA often doesn't apply, but numerous other frameworks do. Consumer health apps face FTC enforcement, state laws, and evolving federal requirements. Key points: - Determine HIPAA applicability correctly - Comply with FTC Health Breach Rule - Address state health privacy laws - Implement proper security measures #### HIPAA applicability limits Contrary to common belief, HIPAA only applies to covered entities (healthcare providers, plans, clearinghouses) and their business associates. Most consumer health apps, fitness trackers, meditation apps, symptom checkers, don't fall under HIPAA because they're not provided by or on behalf of covered entities. However, this doesn't mean health data is unprotected. #### FTC Health Breach Notification Rule Non-HIPAA health apps may trigger the FTC's Health Breach Notification Rule, which requires notification of breaches involving personal health records. The FTC has explicitly clarified this rule applies to health apps. Violations carry civil penalties. Implement breach detection and notification procedures even if HIPAA doesn't apply. #### State health privacy laws Washington's My Health My Data Act and similar state laws create new obligations for consumer health data: consent requirements before collection, limitations on sale of health data, data minimization requirements, and access and deletion rights. These laws often define 'health data' broadly, mood tracking, fitness metrics, and dietary information may qualify. #### FDA medical device considerations Health apps may trigger FDA medical device regulations if they diagnose, treat, or prevent disease. The FDA uses risk-based enforcement discretion, high-risk apps face greater scrutiny. Clinical decision support, diagnostic algorithms, and therapeutic apps are more likely to be regulated. Conduct a proper regulatory classification before launch. ### Ad-funded Mobile Games URL: https://northend.law/guides/ad-funded-mobile-games Section: Website and app legal documents By Wenceslao M. Spiegel. Published 2026-01-28; last reviewed 2026-01-28. General information, not legal advice. Ad-funded mobile games, especially those attracting young audiences, face extensive COPPA regulation. The FTC's 2025 COPPA updates significantly restrict data collection and advertising practices. Key points: - Comply with updated COPPA requirements - Implement proper age verification - Limit data collection from children - Navigate in-app advertising rules #### Updated COPPA requirements The FTC's January 2025 COPPA amendments impose strict new requirements: expanded definition of personal information covering persistent identifiers used for advertising, stricter consent requirements before data collection, limitations on targeted advertising to children, and new data security and retention rules. Games attracting users under 13, even unintentionally, must comply. #### Age verification and gating If your game is directed at children or you have actual knowledge of child users, COPPA applies. Neutral age gates that don't encourage lying about age are permitted, but design choices (characters, content, advertising) may make a game 'directed at children' regardless of age gates. Document your target audience analysis. #### Advertising restrictions COPPA significantly limits advertising to children: no behavioral advertising based on personal information, restrictions on contextual advertising content, limitations on 'push' notifications for monetization, and prohibition on certain persuasive techniques. Ad-funded models for children's games face fundamental constraints, consider alternative monetization. #### Data minimization requirements Under updated COPPA, you can only collect information reasonably necessary for the activity. Collecting device identifiers, location data, or behavioral information beyond immediate game functionality requires verifiable parental consent. Implement technical measures limiting data collection and retention to operational minimums. ### Health data in mobile apps: which US laws apply and what your privacy policy must say URL: https://northend.law/guides/health-data-in-mobile-apps Section: Mobile apps By Wenceslao M. Spiegel. Published 2026-11-05; last reviewed 2026-11-05. General information, not legal advice. For most consumer health and wellness apps, HIPAA does not apply. The data is still regulated: the FTC Act and the FTC's Health Breach Notification Rule apply to apps, Washington, Nevada and Connecticut have consumer health data laws that reach businesses outside those states, and California adds its medical information statute and the CCPA's rules on sensitive personal information. The privacy policy has to match what the app and its partners actually do. ## Does HIPAA apply to my app? Usually not. HIPAA applies to health care providers, health plans and health care clearinghouses (covered entities) and to the businesses that handle protected health information on their behalf (business associates). A fitness tracker, a sleep app, a period tracker or a meditation app that people download for themselves is usually outside it. An app that a hospital or a health plan gives its patients or members, or that you build for one under contract, may be inside it. Being outside HIPAA does not leave the data unregulated. It moves the question to the laws below, several of which are stricter in practice. Source: HHS, covered entities and business associates (https://www.hhs.gov/hipaa/for-professionals/covered-entities/index.html). ## What does the FTC require? Section 5 of the FTC Act prohibits deceptive and unfair practices. The FTC applies it to apps that say one thing in a privacy policy or a store listing and do another in their code, for example by passing health information to an advertising partner through a software development kit (SDK). It has brought cases on that basis against several health and wellness apps, including a prescription discount service, an online counseling service, a fertility app and a period tracker. The orders are listed on the FTC's health privacy page. The FTC also enforces the Health Breach Notification Rule. Its amendments, in effect since 29 July 2024, state that the rule reaches health apps that draw information from more than one source (for example from the user and from a connected device or another app), and that a "breach" includes a disclosure the user did not authorize, not only a security incident. A breach requires notice to the affected users and to the FTC, and in some cases to the media. Sources: 15 U.S.C. § 45 (https://www.law.cornell.edu/uscode/text/15/45); FTC, health privacy (https://www.ftc.gov/business-guidance/privacy-security/health-privacy); FTC, GoodRx Holdings (https://www.ftc.gov/legal-library/browse/cases-proceedings/2023090-goodrx-holdings-inc); FTC, BetterHelp (https://www.ftc.gov/legal-library/browse/cases-proceedings/2023169-betterhelp-inc-matter); FTC, Easy Healthcare Corporation (https://www.ftc.gov/legal-library/browse/cases-proceedings/202-3186-easy-healthcare-corporation-us-v); FTC, Flo Health (https://www.ftc.gov/legal-library/browse/cases-proceedings/192-3133-flo-health-inc); 16 CFR Part 318 (eCFR) (https://www.ecfr.gov/current/title-16/chapter-I/subchapter-C/part-318). ## Which state health data laws apply? Washington. The My Health My Data Act has applied since 31 March 2024 (30 June 2024 for small businesses). It defines consumer health data broadly, including information that identifies a past, present or future health condition and some inferences and location data. It requires separate consent to collect the data and to share it, a signed authorization to sell it, and a consumer health data privacy policy of its own. Individuals can sue under it. A first class action under the Act, about location data collected through SDKs in other companies' apps, was filed in federal court in Seattle in February 2025. Nevada. Nevada's consumer health data law (NRS 603A.400 to 603A.550) follows a similar model. It states that it does not create a private right of action; a violation is treated as a deceptive trade practice. Connecticut. The Connecticut Data Privacy Act treats consumer health data as sensitive data that needs consent, and amendments in Public Act 25-113 widen its scope from 1 July 2026. These laws can apply to an app developer outside those states when it collects data from their residents. Sources: RCW 19.373 (https://app.leg.wa.gov/RCW/default.aspx?cite=19.373); Washington Attorney General, My Health My Data Act (https://www.atg.wa.gov/protecting-washingtonians-personal-health-data-and-privacy); NRS chapter 603A (https://www.leg.state.nv.us/nrs/nrs-603a.html); Connecticut Public Act 25-113 (https://www.cga.ct.gov/2025/act/pa/pdf/2025PA-00113-R00SB-01295-PA.pdf); Connecticut Attorney General, Data Privacy Act (https://portal.ct.gov/ag/sections/privacy/the-connecticut-data-privacy-act). ## What does California add? The Confidentiality of Medical Information Act treats a business that offers consumers software, including a mobile app, designed to maintain medical information as a provider of health care for the purposes of the Act (Civil Code section 56.06). The Act limits disclosure of that information without the user's authorization. The California Consumer Privacy Act treats health information as sensitive personal information. Users may limit its use, and the app must give a notice at collection and publish a privacy policy with set contents. Under the regulations in effect since 1 January 2026, processing sensitive personal information is one of the activities that requires a written risk assessment; for processing that began before that date and continues, the assessment is due by 31 December 2027. CalOPPA requires any operator of an online service, including an app, that collects personal information from California residents to post a privacy policy with set contents. Sources: AB 658 (2013), adding mobile applications to Civil Code section 56.06 (https://www.leginfo.ca.gov/pub/13-14/bill/asm/ab_0651-0700/ab_658_bill_20130827_enrolled.htm); CCPA, Civil Code title 1.81.5 (https://leginfo.legislature.ca.gov/faces/codes_displayText.xhtml?division=3.&part=4.&lawCode=CIV&title=1.81.5); California Privacy Protection Agency, approved regulations (https://cppa.ca.gov/regulations/ccpa_updates.html); Business and Professions Code section 22575 (https://leginfo.legislature.ca.gov/faces/codes_displaySection.xhtml?lawCode=BPC§ionNum=22575). ## What do the app stores require? Apple and Google require developers to declare what data the app and the SDKs it ships with collect, and whether it is used to track the user. Apple's review guidelines also restrict the use of health data for advertising. These declarations are store terms rather than law, but a declaration that does not match the app's behavior can support a deception claim under the FTC Act. Sources: Apple, app privacy details (https://developer.apple.com/app-store/app-privacy-details/); Apple, App Review Guidelines, section 5.1 (https://developer.apple.com/app-store/review/guidelines/#data-collection-and-storage); Google Play, Data safety section (https://support.google.com/googleplay/android-developer/answer/10787469). ## What should the privacy policy say? In plain terms, a health app's privacy policy should: - list the health-related data the app collects, including what it infers; - say why each kind of data is collected and how long it is kept; - name the categories of partners that receive it, including SDK providers, and for what purpose; - explain how users give consent and how they withdraw it; - describe the rights users have and how to use them, by state where the rules differ; - say whether any data is sold or used for advertising, and if so on what basis. Where Washington's law applies, the consumer health data privacy policy is a separate document, and the law requires a prominent link to it. The policy is only as good as its match with the app. A useful first step is to compare three things side by side: what the policy says, what the store labels say, and what the app actually sends to its partners. ## Checklist - HIPAA position recorded: covered entity, business associate, or neither - Every health-related data element listed, including inferences - Every SDK and partner that receives that data identified, with its purpose - Consent flows checked against Washington, Nevada and Connecticut - Privacy policy, store labels and the app's actual data flows compared - Breach procedure covers unauthorized disclosures, not only security incidents - CCPA risk assessment scheduled where sensitive data is processed ## Questions people ask Our app is free and we do not sell data. Do these laws still apply? Yes. Most of them turn on collecting or sharing health data, not on selling it or charging for the app. Passing data to an analytics or advertising SDK can be sharing. We are not in Washington. Does the My Health My Data Act matter? It can. It protects Washington residents and people whose health data is collected in Washington, wherever the business is based. Is a wellness app the same as a health app for these laws? Often, yes. The state laws define health data by what it reveals or suggests about a person's health, not by the name of the app. Do the store labels replace the privacy policy? No. Both are required, and they must tell the same story. ## Update log - 5 November 2026: first published. Sources checked the same day. ### Children and teens in your app: COPPA after April 2026, design codes and app store age signals URL: https://northend.law/guides/children-and-teens-in-apps Section: Mobile apps By Wenceslao M. Spiegel. Published 2026-11-05; last reviewed 2026-11-05. General information, not legal advice. If children under 13 use your app, the amended COPPA Rule applies, and its new duties (separate parental consent before sharing children's data with third parties, a written security program, a written retention policy) had to be in place by 22 April 2026. Teens are covered by state laws, California's Age-Appropriate Design Code (partly enjoined) and a new group of app store laws in Texas, Utah, Louisiana and California that require developers to read the age signal an app store or device sends and to act on it. ## When does COPPA apply to an app? The Children's Online Privacy Protection Act and the FTC's COPPA Rule apply to an app that is directed to children under 13, in whole or in part, and to any app whose operator has actual knowledge that it collects personal information from a child under 13. Whether an app is directed to children depends on its subject, its visual content, its characters, its music, the ages of the people in it and its advertising, among other factors. The operator must give notice to parents, obtain verifiable parental consent before collecting a child's personal information (subject to narrow exceptions), let parents review and delete the data, and keep it secure. Sources: 16 CFR Part 312 (eCFR) (https://www.ecfr.gov/current/title-16/chapter-I/subchapter-C/part-312); FTC, Complying with COPPA: frequently asked questions (https://www.ftc.gov/business-guidance/resources/complying-coppa-frequently-asked-questions). ## What changed in April 2026? The FTC amended the COPPA Rule in a final rule published on 22 April 2025. The amendments took effect on 23 June 2025, and operators had one year from publication, until 22 April 2026, to comply with the changes that carry no earlier date. Among the changes: - Separate consent for sharing. An operator must obtain a separate verifiable parental consent before disclosing a child's personal information to third parties, including for targeted advertising, unless the disclosure is integral to the service. - A written security program. The operator must keep a written information security program suited to the sensitivity of children's data. - A written retention policy. Children's personal information may be kept only as long as reasonably necessary for the purpose it was collected for, and not indefinitely. The policy has to be in writing and published in the online notice. - A wider definition of personal information, which now includes biometric identifiers and government-issued identifiers. In February 2026 the FTC also issued a policy statement on the use of age verification technology under COPPA. Sources: FTC, final rule amendments (https://www.ftc.gov/legal-library/browse/federal-register-notices/16-cfr-part-312-coppa-final-rule-amendments); FTC, press release of January 2025 on the final changes (https://www.ftc.gov/news-events/news/press-releases/2025/01/ftc-finalizes-changes-childrens-privacy-rule-limiting-companies-ability-monetize-kids-data); FTC, policy statement on age verification (February 2026) (https://www.ftc.gov/news-events/news/press-releases/2026/02/ftc-issues-coppa-policy-statement-incentivize-use-age-verification-technologies-protect-children). ## What do state privacy laws say about teens? Several state privacy laws protect older minors as well. Under the California Consumer Privacy Act, a business may not sell or share the personal information of a consumer it knows to be under 16 unless the consumer (if 13 to 15) or a parent (if under 13) has opted in. Connecticut's amendments in Public Act 25-113, in force from 1 July 2026, extend its minors' protections to users under 18 and restrict targeted advertising to minors and the sale of their data. Other states have their own rules, and the thresholds differ. Sources: CCPA, Civil Code section 1798.120 (https://leginfo.legislature.ca.gov/faces/codes_displaySection.xhtml?lawCode=CIV§ionNum=1798.120); Connecticut Public Act 25-113 (https://www.cga.ct.gov/2025/act/pa/pdf/2025PA-00113-R00SB-01295-PA.pdf). ## Where does California's design code stand? California's Age-Appropriate Design Code Act applies to online services, products and features likely to be accessed by children under 18. It has been in court since it was enacted. On 12 March 2026 the Ninth Circuit ruled on the second appeal: - it vacated the preliminary injunction as to the Act as a whole, because the challenge to the Act's coverage definition had not been shown on the record; - it vacated the injunction as to the requirement to estimate users' ages (section 1798.99.31(a)(5)); - it kept the injunction in place for the data use and dark patterns restrictions (section 1798.99.31(b)(1) to (4) and (7)), finding them too vague; - it sent the case back to the district court on whether the rest of the Act can stand without its notice-and-cure provision. The requirement to prepare data protection impact assessments was enjoined in the earlier appeal and remains enjoined. Which parts of the Act can be enforced now depends on what the district court does next, so check the current status before relying on any of it. Sources: Civil Code title 1.81.47 (https://leginfo.legislature.ca.gov/faces/codes_displayText.xhtml?division=3.&part=4.&lawCode=CIV&title=1.81.47); NetChoice, LLC v. Bonta, No. 25-2366 (9th Cir. Mar. 12, 2026) (https://cdn.ca9.uscourts.gov/datastore/opinions/2026/03/12/25-2366.pdf). ## What do the app store age laws ask of developers? A new group of laws puts age checks at the app store or the device, and asks developers to read the result and act on it. - Texas (SB 2420). App stores must verify a user's age category and obtain a parent's consent for minors; developers must use the store's age and consent information and enforce the restrictions it calls for. A federal district court enjoined the Act in December 2025. On 4 June 2026 the Fifth Circuit stayed those injunctions while the appeal proceeds, and on 6 July 2026 the Supreme Court declined to lift the stay. The appeal itself was still pending when this guide was checked. - Utah (SB 142, amended by HB 498 in 2026). The duties of app stores and developers, and the right of minors and parents to sue, begin on 6 May 2027. - Louisiana. The 2025 Act was replaced in 2026 by Act 185 (HB 977), which takes effect on 1 July 2027. - California (AB 1043, the Digital Age Assurance Act). From 1 January 2027, operating systems and app stores must send developers an age-range signal on request. A developer that requests the signal is treated as knowing the user's age range. Apple and Google have released tools that pass an age range to apps where these laws apply. Google's Play Age Signals service, for example, also lets developers tell the store about significant changes that need a parent's approval. Sources: Supreme Court docket No. 25A1389 (https://www.supremecourt.gov/docket/docketfiles/html/public/25a1389.html); Utah HB 498 (2026), enrolled (https://le.utah.gov/Session/2026/bills/enrolled/HB0498.pdf); Louisiana HB 977 (2026), enrolled (https://www.legis.la.gov/legis/ViewDocument.aspx?d=1471343); California AB 1043 (https://leginfo.legislature.ca.gov/faces/billNavClient.xhtml?bill_id=202520260AB1043); Apple, Declared Age Range (https://developer.apple.com/documentation/declaredagerange); Google, Play Age Signals (https://developer.android.com/google/play/age-signals/overview). ## What do the app stores require for children? Google Play's Families policies apply to apps aimed at children: they may use only SDKs that have certified compliance with the Families policies, and they are limited in the identifiers and ads they may use. Apple's Kids category has its own rules on third-party analytics and advertising. Sources: Google Play, Families policies (https://support.google.com/googleplay/android-developer/answer/9893335); Apple, App Review Guidelines (https://developer.apple.com/app-store/review/guidelines/). ## Checklist - Audience assessed: directed to children, mixed audience, or general with actual knowledge - Parental notice and consent flows checked against the amended COPPA Rule - Separate consent obtained before sharing children's data with third parties - Written information security program and written retention policy in place - Teen rules mapped for each state where the app has users - Age signals from the app stores and devices requested, stored and acted on - SDKs checked against the Families policies and the Kids category rules ## Questions people ask Our app is for adults. Can COPPA still apply? Yes, if you have actual knowledge that a user is under 13, for example from an age the user entered or a signal the store sent. Does an age signal from the app store mean we know the user's age? Under California's AB 1043, a developer that requests the signal is treated as knowing the user's age range. Under the Texas and Utah laws, developers must use the store's information. Plan for the signal to change what the law expects of you. Is California's design code in force? In part, and the position is still moving. The Ninth Circuit's March 2026 ruling lifted much of the injunction but kept the data use and dark patterns restrictions blocked, and the district court is considering the rest. Do we need a parent's consent for teens? Not under COPPA, which stops at 13. Several state laws require opt-in consent from teens, or from a parent, for certain uses such as selling their data or targeted advertising. ## Update log - 5 November 2026: first published. Sources checked the same day. ### Do you need a DPA? GDPR Article 28 and CCPA service provider terms URL: https://northend.law/guides/do-you-need-a-dpa Section: Vendor contracts By Wenceslao M. Spiegel. Published 2026-10-16; last reviewed 2026-10-16. General information, not legal advice. Under the GDPR, yes, whenever a vendor processes personal data on your behalf: Article 28(3) requires a written contract with set terms. Under the CCPA, a vendor is a "service provider" or "contractor" only if your written contract contains the terms the statute and the regulations list; without them, handing personal information to the vendor may be treated as a sale or a sharing. ## Who is who? Role | GDPR | CCPA Decides why and how data is processed | Controller (Art. 4(7)) | Business (Civ. Code § 1798.140(d)) Processes data for the other party, on its instructions | Processor (Art. 4(8)) | Service provider (§ 1798.140(ag)) or contractor (§ 1798.140(j)) Receives data and uses it for its own purposes | Another controller | Third party (§ 1798.140(ai)) The label in the contract does not decide the role; what the vendor actually does with the data does. A vendor that uses your customers' data for its own product or advertising is acting for itself, whatever the DPA calls it. ## When does each law require the contract? GDPR. Article 28(3) (https://dposchool.com/gdpr/article/28#3) requires that processing by a processor be governed by a contract (or another legal act) that binds the processor and sets out the subject matter, duration, nature and purpose of the processing, the type of personal data, the categories of data subjects, and the controller's obligations and rights. Article 28(4) (https://dposchool.com/gdpr/article/28#4) requires the same terms to flow down to sub-processors. CCPA. The statute defines a service provider and a contractor by reference to a written contract that contains specific prohibitions (§ 1798.140(ag) and (j)), and requires the business to have that contract (§ 1798.100(d)). The regulations list what the contract must say (Cal. Code Regs., tit. 11, § 7051(a)). A person without a contract that complies with section 7051(a) is not a service provider or contractor, and disclosing personal information to it may be considered a sale or sharing (§ 7050(e)). Sources: GDPR, Regulation (EU) 2016/679 (https://eur-lex.europa.eu/eli/reg/2016/679/oj); CCPA, Civil Code § 1798.100 et seq., as in force on 1 January 2026 (CPPA) (https://cppa.ca.gov/pdf/20260101_ccpa_statute.pdf); CCPA Regulations, effective 1 January 2026 (CPPA) (https://cppa.ca.gov/regulations/pdf/ccpa_statute_eff_20260101.pdf). ## Clause checklist: Article 28(3) and section 7051(a) side by side Subject | GDPR Art. 28(3) | CCPA regulations § 7051(a) Purpose | Processing only on documented instructions, including on transfers (a) | Specific business purposes stated, not in generic terms (2); no use for other purposes (3) Own use by the vendor | Not permitted beyond instructions; a processor that decides purposes becomes a controller (Art. 28(10)) | No selling or sharing (1); no use outside the direct business relationship, and no combining with other data unless permitted (4) Confidentiality | Persons processing the data bound to confidentiality (b) | Not listed as such; covered by the general duty to give the same level of protection (5) Security | Measures required by Article 32 (c) | Reasonable security under Civil Code § 1798.81.5, as an example of the same level of protection (5) Sub-processors | Only with prior authorisation; same terms flow down (d), Art. 28(2) and (4) | Subcontractors must have a compliant contract (§ 7051(b)) Individuals' requests | Assist with data subject rights (e) | Enable the business to comply with consumer requests, or act on the business's instructions (9) Assistance | Assist with security, breach notification, impact assessments and prior consultation (f) | Examples include assisting with cybersecurity audits, risk assessments and automated decision-making duties (5) End of contract | Delete or return the data (g) | Not listed as such; the business's right to stop and remediate (8) Checks | Information and audits, including inspections (h) | Reasonable steps to ensure compliant use, for example reviews, scans, or assessments at least every 12 months (6) Notice of non-compliance | Inform the controller if an instruction infringes the law (Art. 28(3), last paragraph) | Notify the business if it can no longer meet its obligations (7) Sources: GDPR, Article 28 (https://eur-lex.europa.eu/eli/reg/2016/679/oj); CCPA Regulations, §§ 7050 and 7051 (CPPA) (https://cppa.ca.gov/regulations/pdf/ccpa_statute_eff_20260101.pdf). A single DPA can carry both sets of terms. Where the vendor receives data from the EU, the contract will also need a transfer tool (usually the standard contractual clauses, or the vendor's certification under the EU-US Data Privacy Framework); see our guide on EU data transfers after the Data Privacy Framework (/guides/eu-transfers-after-dpf). ## Where do vendor DPAs usually fall short? - Generic purposes. "To provide the services" is the kind of description section 7051(a)(2) does not accept. Name the services. - Product improvement and AI training. The CCPA regulations allow a service provider to use the data internally to build or improve the services it provides to you, but not to perform services for another person (§ 7050(a)(3)). Training a general model that serves all customers needs a close look under both laws. - Sub-processor notice without a real objection right. Article 28(2) requires that you can object to a new sub-processor. A notice that leaves you only the option to terminate, with no refund, should be negotiated. - Audit rights replaced by certificates. A security certificate is useful evidence, but Article 28(3)(h) requires the vendor to allow audits. Agree how they work rather than delete them. - Breach notice in fixed hours, or none. You have 72 hours to notify a supervisory authority under Article 33(1); the processor must notify you without undue delay (Art. 33(2)). Agree a short, fixed period. - Deletion with no date. Set a period for deletion or return at the end of the contract, with a written confirmation. For a clause-by-clause reading of a vendor's standard DPA, see Reviewing a vendor's DPA: the clauses that matter (/guides/reviewing-a-vendor-dpa); for AI tools and APIs, see DPAs with AI vendors: training use, sub-processors and deletion (/guides/dpas-with-ai-vendors). Sources: GDPR, Articles 28 and 33 (https://eur-lex.europa.eu/eli/reg/2016/679/oj); CCPA Regulations, § 7050(a)(3) (https://cppa.ca.gov/regulations/pdf/ccpa_statute_eff_20260101.pdf). ## When should a DPA be reviewed by a lawyer? When the vendor will process sensitive data, children's data or large volumes; when it will use AI models on your data; when data leaves the EU or the UK; when the vendor's paper is "non-negotiable" but your customer contracts promise more than it gives. This firm offers a free sample of its DPA Review Skill, sent by e-mail on request (northend.law/skills (/skills)); the firm's software is built by its affiliated technology company, so we are not a neutral reviewer of it. Need a first draft now? Dealroom, built by our affiliated technology company, generates a DPA under the GDPR and the CCPA. Drafting is free, and we can review the result → (https://dealroom.todo.law/contracts/data-processing-agreement) ## Checklist - Role of the vendor decided on the facts (processor or service provider, or third party) - Purposes stated specifically - Vendor's own use of the data excluded, or expressly limited - Sub-processor list, notice period and objection right - Security measures, breach notice period, assistance duties - Audit or assessment rights - Deletion or return at the end, with a date - Transfer tool where EU or UK data is involved - Same terms flowed down to sub-processors ## Questions people ask Is a DPA required for a US company? If the company processes personal data of people in the EU for an EU customer, or is a controller subject to the GDPR, yes. Under the CCPA, a business needs the service provider terms in its contracts with vendors, whether or not the document is called a DPA. Can we use the vendor's standard DPA? Often, if it contains the terms in the checklist. Read it against both columns of the table; most gaps are in purposes, own use and sub-processors. Is the EU's model Article 28 contract enough? The European Commission's standard contractual clauses between controllers and processors (adopted by an implementing decision of 4 June 2021 under Article 28(7); see the Commission's page (https://commission.europa.eu/publications/standard-contractual-clauses-controllers-and-processors-eueea_en)) meet Article 28, but they do not contain the CCPA terms. Add them. What happens if we have no DPA? Under the GDPR, both parties are in breach of Article 28. Under the CCPA, the vendor is not a service provider, and the disclosure may count as a sale or sharing, which triggers notice and opt-out duties. ## Update log - 16 October 2026: first published; sources checked and citations added the same day. It replaces the earlier, shorter guide on data processing agreements, whose address now leads here. ### Reviewing a vendor's DPA: the clauses that matter URL: https://northend.law/guides/reviewing-a-vendor-dpa Section: Vendor contracts By Wenceslao M. Spiegel. Published 2026-10-29; last reviewed 2026-10-29. General information, not legal advice. This guide is for in-house counsel and privacy leads reviewing a vendor's standard data processing agreement before signing. Whether a DPA is needed at all, and the terms each law requires side by side, are covered in Do you need a DPA? (/guides/do-you-need-a-dpa) For an AI tool or an API, read this guide with DPAs with AI vendors (/guides/dpas-with-ai-vendors). ## Who is really the processor? The label in the contract does not decide the role; what the vendor does with the data does. The controller determines the purposes and means of the processing, and the processor processes personal data on the controller's behalf (GDPR Article 4(7) and (8)). A processor that determines the purposes and means itself is a controller for that processing (Article 28(10) (https://dposchool.com/gdpr/article/28#10)). So a vendor that uses your data for its own purposes (product improvement, analytics, training) is a controller for that use, and the DPA cannot change that by calling it a processor. Where the vendor's terms allow such a use, decide whether you accept it or ask for it to be removed. Under the CCPA, the question is whether the vendor qualifies as a service provider or contractor, or is a third party. A third party receiving personal information from you can make the transfer a sale or sharing. The terms that make a vendor a service provider or contractor are under "The CCPA terms" below. ## Scope and instructions Article 28(3) (https://dposchool.com/gdpr/article/28#3) requires the contract to set out the subject matter and duration of the processing, its nature and purpose, the types of personal data and the categories of data subjects. These details usually sit in an annex. Check that it is filled in for your use, not left generic: it describes what you are authorizing, and it is the measure for the vendor's later use of the data. Under Article 28(3)(a), the vendor acts only on your documented instructions. Watch for "and to improve our services" or similar language that widens the purpose. Each such phrase adds a purpose of the vendor's own, and brings back the question above. ## Sub-processors A processor may engage another processor only with the controller's prior written authorization, specific or general (Article 28(2) (https://dposchool.com/gdpr/article/28#2)). Vendor DPAs almost always ask for general authorization. That is workable only with what Article 28(2) attaches to it: notice of changes and a real right to object. A notice that leaves you only the choice of accepting or terminating should be negotiated. Ask for the current list and the notice period for changes (30 days is common), and require that the same obligations flow down to each sub-processor (Article 28(4) (https://dposchool.com/gdpr/article/28#4)), with the vendor liable to you for them. ## Breach notice You have 72 hours from becoming aware of a breach to notify the supervisory authority (Article 33(1) (https://dposchool.com/gdpr/article/33)); the processor must tell you "without undue delay" (Article 33(2) (https://dposchool.com/gdpr/article/33)). On their own the two do not fit: a vendor that takes three days to tell you leaves you no time. Ask for a fixed window (24 or 48 hours) and a minimum content for the notice (what happened, the data and people affected, the measures taken, a contact), so you can meet your own deadline. ## Audits and assistance Article 28(3)(h) requires the vendor to make available the information needed to demonstrate compliance and to allow audits, including inspections. Vendors usually offer certifications and reports instead (SOC 2, ISO 27001). Accept them as a first step, and keep a right to audit when there is cause, for example after a breach. Article 28(3)(e) and (f) require the vendor to help with data subject requests, security, breach notice and impact assessments. Vendor DPAs often say "at the customer's cost" with no time frame. Check who pays and how fast: help that arrives after your own deadline is of little use. ## Deletion or return at the end Article 28(3)(g) requires the vendor to delete or return all personal data at the end of the service, at your choice, unless the law requires keeping it. Many DPAs repeat the rule without saying when. Ask for a time limit (30 to 90 days), for what happens to backups, and for a written confirmation. ## Liability Standard vendor caps often treat a data breach like any other claim, under the same general cap as a missed service level. Consider a separate, higher cap for data protection breaches, and exclude fines and regulatory costs from the general cap where you can. Read the DPA's liability clause with the main agreement, and check that it does not override the main agreement in the vendor's favor (for example, by bringing claims under the DPA within a lower cap). ## International transfers Where personal data leaves the European Economic Area, Chapter V of the GDPR applies. The transfer needs an adequacy decision (for the United States, the EU-US Data Privacy Framework, which covers only certified US companies), or the Standard Contractual Clauses (Commission Implementing Decision (EU) 2021/914, Module Two, or Module Three where you are yourself a processor) with a transfer impact assessment. Check that sub-processors outside the EEA are covered too: the DPA should say which transfer tool applies to each. See also EU data transfers after the Data Privacy Framework (/guides/eu-transfers-after-dpf). Sources: GDPR, Regulation (EU) 2016/679, EUR-Lex (https://eur-lex.europa.eu/eli/reg/2016/679/oj); Commission Implementing Decision (EU) 2021/914 (Standard Contractual Clauses), EUR-Lex (https://eur-lex.europa.eu/eli/dec_impl/2021/914/oj); Data Privacy Framework program (https://www.dataprivacyframework.gov/). ## The CCPA terms The CCPA requires a written contract with each service provider or contractor (Cal. Civ. Code § 1798.100(d) and § 1798.140), with the terms set out in the CCPA regulations (Cal. Code Regs., tit. 11, § 7051): - the specified business purposes; - no selling or sharing of the personal information; - no use outside the direct business relationship; - no combining with personal information from other sources, except as the regulations permit; - notice to you if the provider can no longer comply; - your right to take reasonable steps to stop and remedy unauthorized use. DPAs written for the GDPR often leave some of these out, and without them the vendor is not a service provider or contractor. One DPA can carry both the GDPR and the CCPA terms. Sources: CCPA, Civil Code § 1798.100 et seq., as in force on 1 January 2026 (CPPA) (https://cppa.ca.gov/pdf/20260101_ccpa_statute.pdf); CCPA Regulations, § 7051 (CPPA) (https://cppa.ca.gov/regulations/pdf/ccpa_statute_eff_20260101.pdf). Need a first draft now? Dealroom, built by our affiliated technology company, generates a DPA under the GDPR and the CCPA. Drafting is free, and we can review the result → (https://dealroom.todo.law/contracts/data-processing-agreement) ## Checklist - Role decided on what the vendor does with the data, not on the label - Vendor's own use of the data (product improvement, analytics, training) removed, or identified and accepted - Annex filled in for your use; processing only on documented instructions, with no wording that widens the purpose - Sub-processor list, notice period and a real right to object; same obligations flowed down, vendor liable for them - Breach notice within a fixed window (24 or 48 hours), with a minimum content - Certifications and reports first, and a right to audit when there is cause - Assistance duties: who pays, and how fast - Deletion or return within a time limit, backups covered, written confirmation - Separate, higher cap for data protection breaches; fines and regulatory costs outside the general cap; no override of the main agreement in the vendor's favor - Transfer tool for the vendor and for each sub-processor outside the EEA - CCPA service provider or contractor terms included ## Questions people ask Can the DPA make the vendor a processor? No. The role follows what the vendor does with the data. A vendor that uses it for its own purposes is a controller for that use, whatever the DPA says. Is a SOC 2 report or an ISO 27001 certificate enough instead of audit rights? It is a reasonable first step, but Article 28(3)(h) requires the vendor to allow audits. Keep a right to audit when there is cause. What breach notice period should we ask for? A fixed window of 24 or 48 hours, with a minimum content, so that you can meet your own 72-hour deadline to notify the supervisory authority. Do we need separate DPAs for the GDPR and the CCPA? No. One DPA can carry both sets of terms. ## Update log - 29 October 2026: first published. ### DPAs with AI vendors: training use, sub-processors and deletion URL: https://northend.law/guides/dpas-with-ai-vendors Section: Vendor contracts By Wenceslao M. Spiegel. Published 2026-10-29; last reviewed 2026-10-29. General information, not legal advice. This guide is for a company buying an AI tool, or an API, that will see personal data in prompts, files or outputs. The usual DPA questions still apply (see Do you need a DPA? (/guides/do-you-need-a-dpa) and Reviewing a vendor's DPA (/guides/reviewing-a-vendor-dpa)); this guide covers the points particular to AI vendors. ## Training on your data The first question is whether the vendor uses your inputs or outputs to train or improve its models. The answer is often in the terms of service or a data usage policy rather than in the DPA, so read them all. If the vendor does train on your data, it is acting as a controller for that use, not as your processor (see Who is really the processor? (/guides/reviewing-a-vendor-dpa#who-is-really-the-processor)), and calling it a processor in the DPA does not change that. Ask for an express prohibition on training, or an opt-in that you control, and check what it covers. A prohibition limited to "customer content" may leave out outputs, uploaded files or feedback (ratings and corrections), which a vendor's terms may treat separately. The clause should name inputs, outputs, files and feedback. ## Model providers as sub-processors Many AI tools do not run their own models: they send your prompts to a model provider's API. That provider is a sub-processor, and the rules on sub-processors apply in full. It must be on the vendor's sub-processor list, with the region where it processes the data, and the vendor's own obligations must flow down to it (Article 28(4) (https://dposchool.com/gdpr/article/28#4)). The flow-down is often the model provider's own business terms; check them for the same points as this guide. Check whether the vendor can switch model providers, or route requests between several. Many can, sometimes by a change of configuration that the customer does not see. A switch to a new provider is the engagement of a new sub-processor, so the DPA should make it trigger the sub-processor notice and your right to object (Article 28(2) (https://dposchool.com/gdpr/article/28#2)). ## Retention of prompts and outputs Ask how long prompts, outputs and logs are kept, by the vendor and by its model provider, including any retention for abuse monitoring, which may apply at the model provider even when the vendor itself keeps nothing. Ask whether a zero-retention option exists, and on what conditions. Logs are personal data too when they contain names or account details, and a vendor may keep them longer than the prompts. The retention terms should cover logs by name. ## Deletion and "unlearning" Deleting data from a database is routine, and Article 28(3) (https://dposchool.com/gdpr/article/28#3), point (g), requires it at the end of the service. Removing the influence of data from a model that has already been trained on it is a different matter: in practice, it is not possible on request. Once data has been used for training, deleting the stored copy does not reach the model. That is why the prohibition on training, above, is the real protection. Ask for deletion of everything stored at the end of the service, within a fixed time: prompts, outputs, uploaded files and embeddings (the numerical representations of your documents that many tools keep for search and retrieval). Ask for a written confirmation, and for the deletion to reach the model provider. ## Transfers to US model providers Where the model provider is in the United States, check whether it is certified under the EU-US Data Privacy Framework; the Framework's website keeps the list. If it is certified, the adequacy decision covers the transfer to it. If it is not, the Standard Contractual Clauses and a transfer impact assessment are needed, as for any other transfer (see International transfers (/guides/reviewing-a-vendor-dpa#international-transfers) and our guide on EU data transfers after the Data Privacy Framework (/guides/eu-transfers-after-dpf)). Ask where processing happens, at the vendor and at the model provider, and whether an EU region can be chosen. A vendor may store data in the EU while its model calls go to the US. Sources: GDPR, Regulation (EU) 2016/679, EUR-Lex (https://eur-lex.europa.eu/eli/reg/2016/679/oj); Data Privacy Framework program (https://www.dataprivacyframework.gov/); Commission Implementing Decision (EU) 2021/914 (Standard Contractual Clauses), EUR-Lex (https://eur-lex.europa.eu/eli/dec_impl/2021/914/oj). ## Access by the vendor's staff Check when the vendor's people can read your prompts and outputs, for example for support when you raise a ticket, or for safety review when a prompt is flagged. Ask under what controls this happens (who may look, on what trigger and with whose approval), and that each access is logged. A vendor that does not train on your data may still have people reading it; the DPA should say when. ## The EU AI Act sits alongside, not inside, the DPA The EU AI Act has its own roles: the provider, who develops an AI system or has it developed and places it on the market, and the deployer, who uses it. These are not the GDPR's roles of controller and processor, and the two sets answer different questions: an AI vendor can be the provider and your processor at once, while you are the deployer and the controller. A deployer's duties (for example, under Article 26 for high-risk AI systems) are handled in the main contract and in your own processes, not in the DPA. The DPA governs the processing of personal data on your behalf. For the dates on which the AI Act's duties apply, see our EU AI Act timeline (/guides/eu-ai-act-timeline). Source: EU AI Act, Regulation (EU) 2024/1689, EUR-Lex (https://eur-lex.europa.eu/eli/reg/2024/1689/oj). ## California: automated decisions The CCPA regulations on automated decision-making technology apply from 2027. If the tool is used to make significant decisions about people, the business using it has notice, access and opt-out duties, and it will need the vendor's help to meet them: for example, information about how the tool reaches its outputs, and a way to act on an opt-out. The contract should support those duties. Ask the vendor what it can provide, and write it into the contract. Source: CCPA Regulations, effective 1 January 2026 (CPPA) (https://cppa.ca.gov/regulations/pdf/ccpa_statute_eff_20260101.pdf). ## What to ask before signing - Do you use our inputs, outputs, files or feedback to train or improve any model? Will you put a prohibition, or an opt-in that we control, in writing? - Which model providers process our data, in which regions, and on what terms? - Can you switch or add model providers, and will a switch trigger the sub-processor notice and our right to object? - How long are prompts, outputs and logs kept, by you and by the model provider, including for abuse monitoring? Is there a zero-retention option? - At the end of the service, will you delete prompts, outputs, files and embeddings within a fixed time, and confirm it in writing? - Is the model provider certified under the EU-US Data Privacy Framework? If not, are the Standard Contractual Clauses in place, with a transfer impact assessment? - Where does processing happen, and can we choose an EU region? - When can your staff read our prompts, under what controls, and is each access logged? - Which of our duties as deployer under the AI Act does the main contract support? - If we use the tool for significant decisions about people, how will you support our notice, access and opt-out duties under the CCPA regulations? ## Questions people ask If the vendor deletes our data, is it also removed from the model? Not if the model has already been trained on it. Deleting stored prompts, outputs, files and embeddings is routine; removing data's influence from a trained model is not, in practice, possible on request. That is why the prohibition on training is the real protection. Is the model provider behind an AI tool a sub-processor? Yes, when the tool sends your data to the provider's API. It must be on the vendor's sub-processor list, with its region, and a switch to another provider should trigger the sub-processor notice. Does the DPA need to cover the EU AI Act? No. The AI Act's roles (provider and deployer) are not the GDPR's roles (controller and processor). A deployer's duties are handled in the main contract and in your own processes. Is a zero-retention option enough on its own? It addresses how long prompts and outputs are kept. It does not settle training use, staff access or transfers, which need their own terms. ## Update log - 29 October 2026: first published. ### Master Service Agreements URL: https://northend.law/guides/master-service-agreements Section: Vendor contracts By Wenceslao M. Spiegel. Published 2026-01-23; last reviewed 2026-01-23. General information, not legal advice. For ongoing client relationships, a well-drafted MSA creates efficiency and reduces friction. It establishes the framework once, letting you focus on specific SOWs for each engagement. Key points: - Create MSA + SOW structure for flexibility - Negotiate liability caps tied to contract value - Define change order and escalation procedures - Include auto-renewal with clear termination rights #### Structure for flexibility The MSA contains your 'evergreen' terms: liability, IP, confidentiality, dispute resolution. Each Statement of Work references the MSA and adds project-specific terms: scope, timeline, pricing. This means you negotiate core terms once and adapt only what changes per project. #### Liability caps should match risk Standard practice ties liability caps to fees paid, often 12 months of fees or the SOW value. Negotiate carve-outs for willful misconduct, IP indemnification, and confidentiality breaches. Ensure caps are mutual. Avoid unlimited liability except for the most egregious conduct. #### Change management prevents disputes Projects evolve. Include a change order process: how changes are requested, how scope impact is assessed, how pricing adjusts, and who has authority to approve. Written change orders prevent 'I thought we agreed...' arguments later. #### Termination rights on both sides Include termination for cause (material breach with cure period) and convenience (with reasonable notice). Define what happens at termination: payment for work completed, deliverable handoff, return of confidential information. Survival clauses for ongoing obligations. ### Scope of Work URL: https://northend.law/guides/scope-of-work Section: Vendor contracts By Wenceslao M. Spiegel. Published 2026-01-23; last reviewed 2026-01-23. General information, not legal advice. The SOW is where abstract agreements meet concrete deliverables. Precision here prevents misaligned expectations, scope creep, and payment disputes. Key points: - List specific deliverables with acceptance criteria - Tie payment milestones to deliverable approval - Include assumptions and exclusions explicitly - Define process for handling out-of-scope requests #### Deliverables need acceptance criteria Don't just list what you'll deliver, define what 'done' means. Specify functionality, performance standards, or quality requirements. Include an acceptance testing period (typically 5-10 business days) and a process for identifying deficiencies that need remediation. #### Tie payments to milestones Structure payments around deliverable acceptance, not just calendar dates. This aligns incentives: you get paid when you deliver value, clients only pay for what meets specifications. Include milestone tables with clear deliverable-to-payment mapping. #### Assumptions and exclusions prevent surprises State what you're assuming: client will provide timely feedback, API access, test data. List what's explicitly excluded: training, ongoing support, third-party integrations. If assumptions prove false, you have grounds to adjust timeline or pricing. #### Handle change requests formally When scope changes (and it will), use a written change request process. Document the requested change, impact analysis, pricing adjustment, and timeline modification. Both parties sign before work proceeds. This protects you from 'while you're at it...' requests. ### Pilot contracts URL: https://northend.law/guides/pilot-contracts Section: Vendor contracts By Wenceslao M. Spiegel. Published 2026-01-23; last reviewed 2026-01-23. General information, not legal advice. Pilots and proof-of-concept engagements are essential for enterprise sales but come with unique legal considerations. Structure them right to protect your interests while removing barriers to conversion. Key points: - Define clear success criteria and timelines - Limit scope to avoid scope creep - Include conversion terms to full contract - Protect your IP during the evaluation period #### Define success criteria upfront Ambiguous pilots lead to endless negotiations. Specify exactly what constitutes a successful pilot: metrics to hit, features to demonstrate, integration milestones. Document these in the agreement. At pilot end, either criteria are met (proceed to full contract) or not met (clear exit). #### Scope limitations prevent scope creep Pilots should test your solution's fit, not customize it for free. Limit users, features, data volumes, or use cases explicitly. If the prospect needs more, that's a signal to move to a paid engagement. Include clear language about what's out of scope. #### Protect your intellectual property Pilots often involve sharing proprietary technology. Include strong confidentiality provisions, prohibit reverse engineering, and clarify that your IP remains yours. If you're building custom integrations, define who owns that work and any derivative insights. #### Build in conversion terms The goal is conversion to a paying customer. Include agreed-upon pricing for the full contract, a timeline for signing decision, and any credits for pilot fees. Make the path from pilot to customer as frictionless as possible while protecting your commercial interests. ### DPIA vs PIA: what each is, when you need one, and what goes in it URL: https://northend.law/guides/dpia-vs-pia Section: Assessments By Wenceslao M. Spiegel. Published 2026-10-16; last reviewed 2026-10-16. General information, not legal advice. A DPIA (data protection impact assessment) is the assessment the GDPR requires before processing that is likely to result in a high risk to people's rights and freedoms (Article 35). "PIA" (privacy impact assessment) is a broader, older term with no fixed legal content of its own; in California, the assessment the law now requires is called a risk assessment, and the CCPA regulations have required one since 1 January 2026 before processing that presents a significant risk to consumers' privacy. A DPIA is the assessment the GDPR requires (Article 35) before processing that is likely to result in a high risk to people's rights. "PIA" is the general term for any privacy impact assessment: no single law defines it, and organizations use it for lower-risk projects or outside the EU. California's risk assessments under the CPPA regulations are a third, separate duty, with their own triggers and filing dates. ## What is a DPIA? Under Article 35(1) of the GDPR, where a type of processing, taking into account its nature, scope, context and purposes, is likely to result in a high risk to the rights and freedoms of natural persons, the controller must carry out a DPIA before the processing. A single assessment may cover a set of similar processing operations that present similar high risks. Article 35(3) names three cases where a DPIA is always required: - a systematic and extensive evaluation of personal aspects based on automated processing, including profiling, on which decisions are based that produce legal or similarly significant effects; - processing on a large scale of special categories of data, or of data relating to criminal convictions and offences; - systematic monitoring of a publicly accessible area on a large scale. Each supervisory authority also publishes a list of the kinds of processing that require a DPIA (Article 35(4)). The European guidelines on DPIAs, endorsed by the European Data Protection Board, give nine criteria: evaluation or scoring; automated decision-making with legal or similar significant effect; systematic monitoring; sensitive or highly personal data; data processed on a large scale; matching or combining datasets; data concerning vulnerable data subjects; innovative use or applying new technological or organisational solutions; and processing that prevents people from exercising a right or using a service or a contract. In most cases, processing that meets two criteria requires a DPIA. Article 35(7) sets the minimum content: a systematic description of the processing and its purposes; an assessment of necessity and proportionality; an assessment of the risks to individuals; and the measures envisaged to address them. Where the DPIA indicates that the processing would result in a high risk in the absence of measures taken by the controller to mitigate it, the controller must consult the supervisory authority before starting (Article 36(1)). Sources: GDPR, Regulation (EU) 2016/679, EUR-Lex (https://eur-lex.europa.eu/eli/reg/2016/679/oj); Article 29 Working Party, Guidelines on DPIA (WP248 rev.01) (https://ec.europa.eu/newsroom/article29/items/611236), endorsed by the EDPB (https://www.edpb.europa.eu/our-work-tools/our-documents/guidelines/data-protection-impact-assessments-high-risk-processing_en). ## What is a PIA? "Privacy impact assessment" is the general term used before the GDPR and still used in many organisations and standards; the European DPIA guidelines themselves refer to the ISO/IEC 29134 work on privacy impact assessment. Used loosely, a PIA is any documented review of a project's privacy risks. A DPIA is the specific, legally defined form required by the GDPR. ## What does California now require? The CCPA regulations (Article 10, sections 7150 to 7157, effective 1 January 2026) require a business whose processing presents significant risk to consumers' privacy to conduct a risk assessment before starting that processing (§ 7150(a)). Six activities present that risk (§ 7150(b)): 1. selling or sharing personal information; 2. processing sensitive personal information (with a narrow exception for certain employment purposes); 3. using automated decisionmaking technology (ADMT) for a significant decision concerning a consumer; 4. using automated processing to infer or extrapolate a person's traits or behaviour from systematic observation of them as an applicant for education or a job, a student, an employee or an independent contractor; 5. using automated processing to do so based on their presence in a sensitive location; 6. processing personal information that the business intends to use to train ADMT for a significant decision, or facial-recognition, emotion-recognition or other technology that verifies identity or identifies or profiles people. The report must document, among other things, the purpose, the categories of personal information, the operational elements, the benefits, the negative impacts on consumers' privacy, the safeguards, whether the business will proceed, who provided information, and the date the assessment was reviewed and approved (§ 7152). The stated goal is to restrict or prohibit processing whose risks to the consumer outweigh its benefits (§ 7154). Dates. For processing that began before 1 January 2026 and continues after that date, the assessment must be done by 31 December 2027 (§ 7155(b)). Assessments must be reviewed at least once every three years, and updated within 45 calendar days of a material change (§ 7155(a)). For assessments conducted in 2026 and 2027, the business must submit summary information and an attestation, made by a member of executive management, to the California Privacy Protection Agency by 1 April 2028 (§ 7157(a), (b)(5) and (c)). The Agency or the Attorney General may ask for the full reports at any time, to be provided within 30 calendar days (§ 7157(e)). Reports are kept for as long as the processing continues or five years after the assessment, whichever is later (§ 7155(c)). Source: CCPA Regulations, effective 1 January 2026 (California Privacy Protection Agency) (https://cppa.ca.gov/regulations/pdf/ccpa_statute_eff_20260101.pdf). ## How do they compare? | GDPR DPIA | California risk assessment Trigger | Processing likely to result in a high risk (Art. 35(1), (3), (4)) | Six listed activities (§ 7150(b)) When | Before the processing | Before the processing; existing processing by 31 December 2027 Content | Description, necessity and proportionality, risks, measures (Art. 35(7)) | Purpose, data, operations, benefits, negative impacts, safeguards, decision, contributors, approval (§ 7152) Who is involved | Advice of the DPO, where designated (Art. 35(2)); views of data subjects where appropriate (Art. 35(9)) | Employees who take part in the processing; external parties may be included (§ 7151) Review | Where necessary, at least when the risk changes (Art. 35(11)) | At least every three years; within 45 days of a material change (§ 7155) Regulator | Prior consultation where the risk would be high without mitigating measures (Art. 36) | Submission of summary information and an attestation (§ 7157) ## Can one document serve both? Yes, if it contains everything each law asks for. The California regulations allow a business to use an assessment prepared for another purpose, provided it contains, or is paired with, the information section 7152 requires (§ 7156(b)). In practice, add to a GDPR DPIA the California fields it usually lacks: the operational elements, the benefits, the names of contributors, the decision whether to proceed, and the approval. Under the EU AI Act, some deployers of high-risk AI systems must also carry out a fundamental rights impact assessment (Article 27); see our EU AI Act timeline (/guides/eu-ai-act-timeline). ## Checklist - Screen every new project against Article 35 (and the national list) and against § 7150(b) - Where a listed case or criterion applies but you decide no DPIA is needed, record the reasons and the DPO's views - Use one template that covers Article 35(7) and § 7152 - Involve the DPO (EU) and the employees and external parties § 7151 describes (California) - Finish before the processing starts - List all processing begun before 2026 that § 7150(b) covers; assess it by 31 December 2027 - Diary the 1 April 2028 submission and name the executive who will attest - Set review dates: at least every three years, and within 45 days of a material change ## Questions people ask Is a PIA the same as a DPIA? Not quite. A DPIA is the GDPR's defined assessment; a PIA is a general term. A well-built PIA can meet the DPIA requirements if it covers Article 35(7). Do US companies need a DPIA? Only where the GDPR applies to the processing. Under the CCPA, a business needs a risk assessment for the activities in § 7150(b), whether or not it also needs a DPIA. What happens if we skip it? Under the GDPR, the obligations in Articles 25 to 39, which include Article 35, are subject to fines of up to 10 million euros or, for an undertaking, up to 2% of its total worldwide annual turnover of the preceding financial year, whichever is higher (Article 83(4)). Under the CCPA, the submission to the Agency includes an attestation made under penalty of perjury (§ 7157(b)(5)). Who signs? Under the GDPR, the controller carries out the DPIA and the DPO advises. In California, the report records the date it was reviewed and approved, and a member of executive management makes the submission to the Agency. Sources: GDPR, Article 83, EUR-Lex (https://eur-lex.europa.eu/eli/reg/2016/679/oj); CCPA Regulations, § 7157 (CPPA) (https://cppa.ca.gov/regulations/pdf/ccpa_statute_eff_20260101.pdf). ## Update log - 16 October 2026: first published; sources checked and citations added the same day. ### How to run a DPIA in practice, with a checklist URL: https://northend.law/guides/how-to-run-a-dpia Section: Assessments By Wenceslao M. Spiegel. Published 2026-10-16; last reviewed 2026-10-16. General information, not legal advice. Run a DPIA before the processing starts, in nine steps: screen the project, describe the processing, test necessity and proportionality, identify the risks to people, choose measures, take the DPO's advice, decide on the remaining risk (consulting the supervisory authority where required), build the measures into the project, and review the assessment when the risk changes. Article 35(7) of the GDPR sets the minimum content; the steps below are a practical way to get there. ## Step 1. Does the project need a DPIA? Check the project against Article 35(3), against the list published by the competent supervisory authority (Article 35(4)), and against the nine criteria in the European DPIA guidelines (in most cases, processing that meets two of them needs a DPIA). If the processing matches those cases but you conclude that it is not likely to result in a high risk, the guidelines say to justify and document the reasons, and to record the views of the data protection officer. If the project also falls under § 7150(b) of the California regulations, plan one document that serves both laws (see DPIA vs PIA (/guides/dpia-vs-pia)). ## Step 2. How do you describe the processing? Article 35(7)(a) asks for a systematic description of the processing operations and their purposes, including, where applicable, the legitimate interest pursued. Cover: what data, about whom, from which sources; what is done with it and by which systems; who receives it (including processors and their sub-processors); where it is stored and transferred; and how long it is kept. A data-flow diagram saves pages. ## Step 3. Is the processing necessary and proportionate? Article 35(7)(b). For each purpose, state the legal basis, why each data item is needed, the retention period, how people are informed, and how they can exercise their rights. Where a less intrusive way would achieve the purpose, say why it was not chosen. ## Step 4. What are the risks to people? Article 35(7)(c) is about the risks to the rights and freedoms of the people concerned, not risks to the company. For each risk (for example unlawful access, loss, unexpected use, discrimination, or loss of control over the data), record how likely it is and how severe it would be, on a scale used the same way throughout. ## Step 5. Which measures reduce them? Article 35(7)(d): the measures envisaged to address the risks, including safeguards, security measures and mechanisms to protect the data and to demonstrate compliance. Examples are pseudonymisation, access controls, shorter retention, encryption, contractual limits on processors (Article 28), human review of automated decisions, and clear notices. Record the risk that remains after each measure. ## Step 6. Who must be consulted? - The DPO, where one is designated: the controller must seek the DPO's advice (Article 35(2)). Record the advice. - The people concerned or their representatives, where appropriate (Article 35(9)). - Processors, who must assist the controller with the obligations in Articles 32 to 36 (Article 28(3)(f)). - In California, the employees who take part in the processing must be included, and external parties may be (§ 7151 of the CCPA regulations). ## Step 7. What if the risk stays high? The controller decides whether to proceed. Where the DPIA indicates that the processing would result in a high risk in the absence of measures taken by the controller to mitigate it, the controller must consult the supervisory authority before starting (Article 36(1)). If the authority considers that the processing would infringe the GDPR, it gives written advice within up to eight weeks, which may be extended by six weeks for complex processing (Article 36(2)). ## Step 8. How do the measures reach the project? Turn each measure into a task with an owner and a date: a configuration, a contract clause, a notice, a training. A DPIA whose measures never reach the build does not protect anyone. ## Step 9. When is it reviewed? Where necessary, the controller must review whether the processing follows the DPIA, at least when the risk represented by the processing changes (Article 35(11)). Set a review date and a list of triggers: new data, a new purpose, a new vendor, a new country, a new technology. In California, a risk assessment is reviewed at least once every three years and updated within 45 calendar days of a material change (§ 7155(a)). Sources: GDPR, Articles 28, 35 and 36, EUR-Lex (https://eur-lex.europa.eu/eli/reg/2016/679/oj); Guidelines on DPIA (WP248 rev.01) (https://ec.europa.eu/newsroom/article29/items/611236); CCPA Regulations, §§ 7151 and 7155 (https://cppa.ca.gov/regulations/pdf/ccpa_statute_eff_20260101.pdf). ## Checklist - Screening recorded (Article 35(3) and (4), the nine criteria, § 7150(b)), with reasons where no DPIA is done - Systematic description, with a data-flow diagram - Legal basis, data needed, retention, notices and rights for each purpose - Risks to people, each with likelihood and severity - Measures for each risk, and the risk that remains - DPO advice recorded; people concerned consulted where appropriate; processors' input - Decision to proceed recorded; prior consultation where Article 36(1) requires it - Measures assigned to owners with dates - Review date and triggers set ## Questions people ask How long does a DPIA take? The law sets no duration; it depends on the scope. In our view, a single product feature with known vendors needs far less work than a new platform with many data sources, and starting at the design stage is quicker than assessing a finished system. Can we use a template? Yes. The European DPIA guidelines list methodologies published by several national authorities (Annex 1 of the guidelines). A template helps only if the description in step 2 is specific to the project. Does a vendor's DPIA cover us? No. The controller carries out the DPIA. A vendor's own assessment, security documents and answers are inputs to yours, and a processor must assist (Article 28(3)(f)). Do we publish the DPIA? The GDPR does not require it. The European DPIA guidelines suggest that controllers consider publishing at least parts of it, such as a summary or the conclusion. ## Update log - 16 October 2026: first published; sources checked and citations added the same day. ### EU AI Act timeline after the omnibus: what applies when URL: https://northend.law/guides/eu-ai-act-timeline Section: EU AI Act and US AI rules By Wenceslao M. Spiegel. Published 2026-10-16; last reviewed 2026-10-16. General information, not legal advice. The prohibitions, the AI literacy duty and the rules for general-purpose AI models already apply, and the transparency duties of Article 50 apply from 2 August 2026. The 2026 omnibus moved the start of the high-risk rules to 2 December 2027 for the uses listed in Annex III, and to 2 August 2028 for AI in products covered by Annex I. The EU AI Act is Regulation (EU) 2024/1689. It was amended in 2026 by the Digital Omnibus on AI, Regulation (EU) 2026/1744, which was published in the Official Journal on 24 July 2026 and entered into force on 27 July 2026. ## What applies when? Date | What applies 1 August 2024 | The AI Act enters into force 2 February 2025 | Prohibited AI practices and AI literacy obligations 2 August 2025 | Governance rules and the obligations for general-purpose AI models 2 August 2026 | General date of application, including the transparency duties of Article 50 2 December 2026 | Marking of AI-generated content (Article 50(2)) for generative systems placed on the market before 2 August 2026; the new prohibition of AI systems that generate non-consensual intimate material or child sexual abuse material 2 December 2027 | High-risk uses listed in Annex III (for example recruitment software, credit scoring, scoring of exams) 2 August 2028 | High-risk AI systems in products covered by Annex I (Article 6(1)) Sources: European Commission, AI Act page (application timeline) (https://digital-strategy.ec.europa.eu/en/policies/regulatory-framework-ai); European Parliament, Legislative Observatory, summary of Regulation (EU) 2026/1744 (https://oeil.europarl.europa.eu/oeil/en/document-summary?id=1910754); European Commission, transparency obligations under Article 50 (https://digital-strategy.ec.europa.eu/en/faqs/transparency-obligations-under-article-50-ai-act); Regulation (EU) 2024/1689, EUR-Lex (https://eur-lex.europa.eu/eli/reg/2024/1689/oj); Regulation (EU) 2026/1744, EUR-Lex (https://eur-lex.europa.eu/eli/reg/2026/1744/oj/eng). ## What moved, and what did not? Moved. The start dates of the main high-risk requirements, and a short extra period for marking AI-generated content made by systems already on the market. The omnibus also extends several simplified compliance measures for small and medium-sized enterprises to a new category of small mid-cap enterprises, and adds a prohibition, covering non-consensual intimate material and child sexual abuse material, that applies from 2 December 2026. Did not move. The prohibitions already in force, the rules for general-purpose models, and Article 50, which still applies from 2 August 2026. AI literacy was reworded, not removed. Providers and deployers must take measures to support AI literacy among their staff and the persons operating AI systems on their behalf. They are no longer required to guarantee a particular level of AI literacy for each individual. The Commission and the Member States must support these efforts, particularly for smaller companies. Sources: European Parliament, Legislative Observatory, summary of Regulation (EU) 2026/1744 (https://oeil.europarl.europa.eu/oeil/en/document-summary?id=1910754); European Commission, AI Act page (https://digital-strategy.ec.europa.eu/en/policies/regulatory-framework-ai). ## Does it apply to a US company? Often, yes. The Act applies to providers placing AI systems or general-purpose AI models on the market in the EU, "irrespective of whether those providers are established or located within the Union or in a third country" (Article 2(1)(a)), and to providers and deployers established outside the EU "where the output produced by the AI system is used in the Union" (Article 2(1)(c)). A US company with EU customers should assume it is in scope until it has checked. Source: AI Act Service Desk, Article 2 (https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-2). ## What should a US company do in each period? Now. - Confirm that no product uses a prohibited practice, including the new one that applies from 2 December 2026. - Keep a record of the measures taken on AI literacy (training, guidance, who received it). - If you provide a general-purpose model, check the rules for providers of such models, including those on transparency and copyright. - Check Article 50: AI systems that interact directly with people, such as chatbots, AI agents and avatars, must be designed so that people are informed they are interacting with AI; generative systems must mark their output (by 2 December 2026 for systems already on the market before 2 August 2026). See the Commission's questions and answers on Article 50 (https://digital-strategy.ec.europa.eu/en/faqs/transparency-obligations-under-article-50-ai-act). Before 2 December 2027. - Classify each AI system against Annex III. For each high-risk system, plan the risk management system, data governance, technical documentation, logging, human oversight and the conformity assessment. - Some deployers of high-risk systems, including bodies governed by public law and private entities providing public services, must carry out a fundamental rights impact assessment (Article 27). Before 2 August 2028. - If your AI is a product, or a safety component of a product, covered by the EU product laws in Annex I, align the AI Act requirements with that product's conformity assessment. Source: AI Act Service Desk, Article 27 (https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-27). ## Checklist - Inventory of AI systems and models, with the role you hold for each (provider, deployer, importer, distributor) - Prohibited practices excluded, including the new prohibition - AI literacy measures recorded - Article 50 disclosures in place; marking of generated content scheduled - Annex III classification done; plan for 2 December 2027 - Annex I products identified; plan for 2 August 2028 ## Questions people ask Was the AI Act delayed? In part. The main high-risk requirements start later (2 December 2027 and 2 August 2028). The other dates did not move. Do the transparency rules apply now? Yes, Article 50 applies from 2 August 2026. Generative systems already on the market before that date have until 2 December 2026 to mark their output. Is AI literacy still required? Yes. Providers and deployers must take measures to support it, but need not guarantee a particular level for each person. Do we need to do anything about high-risk systems before December 2027? The requirements do not apply before then, but the work (classification, documentation, testing) takes months, and contracts with EU customers often ask for it earlier. ## Update log - 16 October 2026: first published, with the dates of Regulation (EU) 2026/1744. This guide is reviewed monthly. Sources checked the same day; the new prohibition is described as one prohibition, as the Commission counts it. ### AI compliance URL: https://northend.law/guides/ai-compliance Section: EU AI Act and US AI rules By Wenceslao M. Spiegel. Published 2026-01-23; last reviewed 2026-01-23. General information, not legal advice. AI regulation is rapidly evolving, with the EU AI Act, Colorado AI Act, and other frameworks creating new compliance obligations. Proactive compliance today prevents costly remediation tomorrow. Key points: - Disclose AI usage in customer-facing products - Document training data sources and licensing - Implement human oversight for automated decisions - Monitor EU AI Act and state-level requirements #### Transparency is foundational Users have a right to know when they're interacting with AI. Disclose AI involvement in customer-facing features, especially for content generation, recommendations, or decisions affecting users. Be specific about what the AI does and doesn't do. #### Training data and model documentation Know your supply chain. Document the provenance of training data, licensing terms for any third-party datasets, and potential biases in your models. If using foundation models, understand their training data and your indemnification rights for IP claims. #### High-risk use cases need human oversight Automated decisions affecting employment, credit, insurance, housing, or education face heightened scrutiny. Implement human review for consequential decisions, document your decision logic, and provide mechanisms for users to contest automated outcomes. #### Stay current with evolving requirements The EU AI Act phases in through 2027 with different requirements for different risk levels. US states are passing their own AI laws. Monitor developments through legal counsel, industry groups, and regulatory announcements. Build compliance infrastructure that can adapt. ### Customer-facing Chatbots URL: https://northend.law/guides/customer-facing-chatbots Section: EU AI Act and US AI rules By Wenceslao M. Spiegel. Published 2026-01-28; last reviewed 2026-01-28. General information, not legal advice. AI chatbots handling customer interactions face specific regulatory requirements for transparency, accuracy, and consumer protection. The FTC has launched targeted investigations into AI chatbot practices. Key points: - Disclose AI nature to users clearly - Avoid deceptive practices per FTC guidance - Handle sensitive data appropriately - Implement human escalation pathways #### FTC focus on AI chatbots The FTC launched an inquiry into AI chatbots in September 2025, focusing on companion and customer service applications. Key concerns include: inadequate disclosure that users are interacting with AI, deceptive claims about chatbot capabilities, data collection practices from conversations, and harms from chatbot advice or recommendations. Expect enforcement to follow investigation. #### Disclosure requirements Both FTC guidance and EU AI Act require clear disclosure when users interact with AI. Disclosure must be: timely (before or at interaction start), clear (not buried in terms of service), and conspicuous (users actually see it). Simple statements like 'You're chatting with an AI assistant' suffice, but must be genuinely visible in the interface. #### Avoiding deceptive practices Don't claim your chatbot can do things it can't. Common pitfalls: overstating accuracy of responses, implying human-level understanding, providing advice in regulated domains (medical, legal, financial) without appropriate disclaimers, and failing to disclose limitations. Test your chatbot's responses in edge cases and document known limitations. #### Data handling in conversations Chatbot conversations may contain sensitive information users didn't intend to share. Implement: data minimization (don't store more than necessary), clear retention policies, redaction of sensitive data from training, and privacy-preserving conversation logging. Consider whether conversations are covered by your privacy policy's disclosures. ### AI Assistants URL: https://northend.law/guides/ai-assistants Section: EU AI Act and US AI rules By Wenceslao M. Spiegel. Published 2026-01-28; last reviewed 2026-01-28. General information, not legal advice. Enterprise AI assistants, from coding copilots to document analysis tools, create new categories of legal risk. Proper governance frameworks address confidentiality, accuracy, and liability allocation. Key points: - Establish AI acceptable use policies - Manage confidential data in AI systems - Document AI decision-making processes - Allocate liability for AI outputs #### Confidential data in AI systems Employees often input confidential information into AI assistants without considering implications. Risks include: exposure of trade secrets through training data, violation of client confidentiality obligations, inadvertent waiver of attorney-client privilege, and breach of contractual confidentiality terms. Implement technical controls preventing sensitive data input and train employees on appropriate use. #### AI acceptable use policies Every organization using AI assistants needs clear policies covering: prohibited use cases (confidential client data, regulated decisions), required human review for outputs, citation and verification requirements, data handling and retention, and accountability for AI-assisted work. Update policies as AI capabilities evolve. #### Liability for AI outputs When AI-assisted work causes harm, who's liable? Generally, the human using the tool remains responsible, AI doesn't create immunity. Professionals retain their duties of care regardless of AI assistance. Contracts should address: disclaimers about AI use, representations about output accuracy, and indemnification for AI-related errors. #### Documentation and audit trails Maintain records of AI involvement in significant decisions: which tools were used, what inputs were provided, what outputs were generated, and what human review occurred. This documentation supports defensibility if decisions are challenged and enables consistent quality oversight of AI-assisted processes. ### Brain Computer Interfaces URL: https://northend.law/guides/brain-computer-interfaces Section: EU AI Act and US AI rules By Wenceslao M. Spiegel. Published 2026-01-28; last reviewed 2026-01-28. General information, not legal advice. Brain-Computer Interfaces represent a new frontier of privacy regulation. Neural data, information derived from brain activity, is receiving dedicated legal protection as legislators recognize its sensitivity. Key points: - Understand emerging neural data laws - Implement heightened consent requirements - Address FDA medical device regulations - Prepare for the MIND Act and state laws #### Emerging neural data legislation Colorado and California amended their privacy laws in 2024 to explicitly protect neural data. Colorado's law includes neural data in its definition of sensitive personal information requiring heightened consent. California's amendment added neural data to biometric information protections. The proposed federal MIND Act would establish national standards. More states are considering similar measures. #### Heightened consent requirements Neural data laws typically require express consent before collection, not just disclosure and opt-out. Given the intimate nature of brain activity data, standard privacy policy consent mechanisms may be insufficient. Consider dedicated consent flows specifically addressing: what neural data is collected, how it's processed, whether it's shared, and how long it's retained. #### FDA regulatory considerations Many BCIs qualify as medical devices subject to FDA regulation. Classification depends on intended use: devices claiming to diagnose or treat conditions face stricter requirements than 'wellness' products. The FDA has issued guidance on non-invasive BCI devices. Regulatory strategy should be determined early, FDA compliance affects product design, clinical testing, and market authorization. #### Ethical considerations beyond compliance Neural data raises profound ethical questions that law hasn't fully addressed: mental privacy (right to cognitive liberty), neural rights (protection from unwanted brain manipulation), data sovereignty (who owns insights from your thoughts). Companies in this space should anticipate evolving ethical expectations and build trust through transparency and restraint. ### Transferring personal data out of the EU after the Data Privacy Framework: SCCs, TIAs and the US adequacy decision URL: https://northend.law/guides/eu-transfers-after-dpf Section: International data transfers By Wenceslao M. Spiegel. Published 2026-10-16; last reviewed 2026-10-16. General information, not legal advice. Personal data can flow from the EU to a US company that participates in the EU-US Data Privacy Framework on the basis of the European Commission's adequacy decision of 10 July 2023, without standard contractual clauses or a transfer impact assessment for that transfer. For a US recipient that does not participate, and for most other countries, use the standard contractual clauses and document a transfer impact assessment. ## How does the GDPR control transfers? Chapter V of the GDPR (Articles 44 to 49) allows a transfer of personal data to a country outside the European Economic Area only on one of three grounds: 1. an adequacy decision of the Commission for that country, territory or sector (Article 45); 2. appropriate safeguards, most often the Commission's standard contractual clauses, or binding corporate rules (Article 46); 3. a derogation for a specific situation, such as explicit consent or necessity for a contract with the individual (Article 49). Derogations are for specific situations, not for routine flows. Source: GDPR, Chapter V (Articles 44 to 49), EUR-Lex (https://eur-lex.europa.eu/eli/reg/2016/679/oj). ## What does the Data Privacy Framework cover? The Commission's adequacy decision for the EU-US Data Privacy Framework, adopted on 10 July 2023 (Implementing Decision (EU) 2023/1795), covers commercial organisations in the United States that participate in the Framework. The protections it relies on include Executive Order 14086 of 7 October 2022, "Enhancing Safeguards for United States Signals Intelligence Activities". What to check before relying on it: - The recipient participates. The US Department of Commerce publishes the list of participating organisations at dataprivacyframework.gov. Check the name, the status and the date. - The data is covered. Participation is for stated types of data; human resources data needs HR coverage. - Onward transfers. The recipient's own transfers to its vendors must follow the Framework's onward transfer principle; your contract should say so. Sources: European Commission, adequacy decisions (https://commission.europa.eu/law/law-topic/data-protection/international-dimension-data-protection/adequacy-decisions_en); Implementing Decision (EU) 2023/1795, with the Framework's Principles in Annex I (including Supplemental Principle 9 on human resources data), EUR-Lex (https://eur-lex.europa.eu/eli/dec_impl/2023/1795/oj); Executive Order 14086, Federal Register (https://www.federalregister.gov/documents/2022/10/14/2022-22531/enhancing-safeguards-for-united-states-signals-intelligence-activities); Data Privacy Framework list (https://www.dataprivacyframework.gov/list). ## Is the Framework safe to rely on? It is in force. On 3 September 2025, the General Court dismissed an action for annulment of the adequacy decision (Case T-553/23, Latombe v Commission), confirming that, on the date the decision was adopted, the United States ensured an adequate level of protection for personal data transferred to organisations under the Framework. A judgment of the General Court can be appealed to the Court of Justice on points of law; check the Court's register for any appeal before relying on the Framework for a new data flow. The previous arrangement, the Privacy Shield, was invalidated by the Court of Justice in the Schrems II judgment of 16 July 2020. A prudent exporter therefore keeps a fallback: standard contractual clauses signed with its main US recipients, ready to rely on if the decision falls. Sources: Court of Justice of the EU, press release No 106/25 (https://curia.europa.eu/site/upload/docs/application/pdf/2025-09/cp250106en.pdf); press release No 91/20 (Schrems II) (https://curia.europa.eu/jcms/upload/docs/application/pdf/2020-07/cp200091en.pdf). ## When do you need standard contractual clauses? When the recipient is outside the EEA and not subject to the GDPR, no adequacy decision covers it, and no binding corporate rules apply. The Commission issued modernised standard contractual clauses on 4 June 2021, with four modules: Module | Exporter | Importer 1 | Controller | Controller 2 | Controller | Processor 3 | Processor | Processor 4 | Processor | Controller The parties may supplement the clauses with additional clauses or incorporate them into a broader commercial contract, as long as nothing in the other terms contradicts the clauses. The clauses are designed for importers outside the EEA that are not themselves subject to the GDPR. Sources: Implementing Decision (EU) 2021/914, EUR-Lex (https://eur-lex.europa.eu/eli/dec_impl/2021/914/oj); European Commission, standard contractual clauses (https://commission.europa.eu/law/law-topic/data-protection/international-dimension-data-protection/standard-contractual-clauses-scc_en); Commission questions and answers on the clauses (https://commission.europa.eu/law/law-topic/data-protection/international-dimension-data-protection/new-standard-contractual-clauses-questions-and-answers-overview_en). ## What is a transfer impact assessment, and when do you need one? In line with the Schrems II judgment, Clause 14 of the standard contractual clauses requires the parties to assess, before concluding the clauses, whether the laws and practices of the destination country could prevent the importer from complying with them, and to document that assessment. The European Data Protection Board's Recommendations 01/2020, adopted in final form on 18 June 2021, describe how to do it: know your transfers; identify the transfer tool; assess the destination's law and practice; adopt supplementary measures where needed; take any formal steps; and re-evaluate at intervals. For a US recipient that does not participate in the Framework, the assessment can take account of the US safeguards the Commission examined in its adequacy decision, including Executive Order 14086: the Commission states that these safeguards apply to all transfers to companies in the US under the GDPR, whatever transfer tool is used. For other countries, the assessment has to be done country by country. Sources: EDPB, Recommendations 01/2020 (https://www.edpb.europa.eu/our-work-tools/our-documents/recommendations/recommendations-012020-measures-supplement-transfer_en); European Commission, questions and answers on the EU-US Data Privacy Framework (question 7) (https://ec.europa.eu/commission/presscorner/detail/en/qanda_23_3752). ## What about transfers to AI providers? Sending personal data to a model API or an inference host outside the EEA is a transfer like any other. Check whether the provider participates in the Framework (and for which data), which entity you contract with, where inference and logging take place, which sub-processors it uses, and whether it may use your data to train its models. The data processing agreement should answer each point (see Do you need a DPA? (/guides/do-you-need-a-dpa)). ## Checklist - Map every transfer: data, recipient, country, purpose, tool relied on - For US recipients: participation checked on the official list, with the date and the data covered - For everyone else: standard contractual clauses, with the right module - Transfer impact assessment documented (Clause 14), with supplementary measures where needed - Onward transfers and sub-processors covered in the contract - Fallback clauses in place for key US recipients that rely on the Framework - Review date set, and a trigger for any court ruling on the Framework ## Questions people ask Do we need SCCs if the US recipient participates in the Framework? Not for the transfer to that recipient, as long as the adequacy decision stands and the participation covers the data. Many companies sign them anyway as a fallback. Do we need a TIA for a participating US recipient? Not for transfers covered by the adequacy decision. A TIA is needed when you rely on the standard contractual clauses. What if the decision is annulled? Transfers relying only on the Framework would lose their basis. Standard contractual clauses already signed, with a documented assessment, would allow transfers to continue while the situation is assessed. Is consent a good basis for transfers? Rarely. Explicit consent under Article 49(1)(a) is for specific situations, and it can be withdrawn. ## Update log - 16 October 2026: first published; sources checked and citations added the same day. ### UK and Swiss transfers: the IDTA, the UK Addendum and the Swiss list URL: https://northend.law/guides/uk-and-swiss-transfers Section: International data transfers By Wenceslao M. Spiegel. Published 2026-10-16; last reviewed 2026-10-16. General information, not legal advice. The United Kingdom and Switzerland each have their own transfer rules. From the UK, rely on UK adequacy regulations (including those for US companies in the UK Extension to the Data Privacy Framework) or, failing that, the International Data Transfer Agreement (IDTA) or the UK Addendum to the EU standard contractual clauses; from Switzerland, check the Federal Council's list of countries with adequate protection (which includes US companies certified under the Swiss-US Data Privacy Framework) or use the EU clauses with the Swiss adaptations. ## How does the UK control transfers? The UK GDPR allows a restricted transfer under UK adequacy regulations, under appropriate safeguards, or under an exception. The Data (Use and Access) Act 2025 changed the test: the level of protection in the destination must be not materially lower than in the UK (the "data protection test"). Section 85 of and Schedule 7 to the Act, which make this change, came into force on 5 February 2026 under the Data (Use and Access) Act 2025 (Commencement No. 6 and Transitional and Saving Provisions) Regulations 2026 (SI 2026/82). The ICO publishes guidance on international transfers, including how to complete a transfer risk assessment. Sources: Data (Use and Access) Act 2025, section 85 (https://www.legislation.gov.uk/ukpga/2025/18/section/85) and Schedule 7 (https://www.legislation.gov.uk/ukpga/2025/18/schedule/7); SI 2026/82, legislation.gov.uk (https://www.legislation.gov.uk/uksi/2026/82/made); ICO, international transfers (https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-resources/international-transfers/). ## What are the IDTA and the UK Addendum? Both are UK standard data protection clauses, of the kind provided for in section 119A of the Data Protection Act 2018. - The IDTA is a stand-alone UK agreement for restricted transfers. - The UK Addendum is added to the EU standard contractual clauses, so that one set of clauses covers EU and UK transfers together. For groups that already use the EU clauses, it is usually the simpler choice. The ICO's guidance says that if you plan to use one of the Article 46 safeguards, which include the IDTA and the Addendum, you must first complete a transfer risk assessment (now called the "data protection test" in the legislation). Sources: Data Protection Act 2018, section 119A (https://www.legislation.gov.uk/ukpga/2018/12/section/119A); ICO, appropriate safeguards (https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-resources/international-transfers/appropriate-safeguards/); ICO, completing a transfer risk assessment (https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-resources/international-transfers/completing-a-transfer-risk-assessment/). ## What is the UK Extension for US companies? The Data Protection (Adequacy) (United States of America) Regulations 2023 came into force on 12 October 2023. They cover transfers from the UK to US organisations that participate in the UK Extension to the EU-US Data Privacy Framework. Check on the official list that the recipient has joined the UK Extension, not only the EU Framework. Sources: SI 2023/1028, legislation.gov.uk (https://www.legislation.gov.uk/uksi/2023/1028/made); Data Privacy Framework list (https://www.dataprivacyframework.gov/list). ## What about transfers from the EU to the UK? The European Commission renewed its adequacy decisions for the United Kingdom in December 2025, under the GDPR and under the Law Enforcement Directive. Personal data can therefore flow from the EU to the UK without extra safeguards. Source: European Commission, adequacy decisions (https://commission.europa.eu/law/law-topic/data-protection/international-dimension-data-protection/adequacy-decisions_en). ## How does Switzerland control transfers? Under the Federal Act on Data Protection of 25 September 2020 (FADP), personal data may be disclosed abroad if the Federal Council has decided that the destination guarantees an adequate level of protection. Annex 1 to the Data Protection Ordinance lists those countries; all EU and EEA countries are on it. Otherwise, appropriate safeguards are needed, such as standard data protection clauses (Article 16(2)(d) FADP). - United States. From 15 September 2024, the list includes the United States for personal data processed by companies certified under the Swiss-US Data Privacy Framework. - The EU clauses. The Federal Data Protection and Information Commissioner (FDPIC) accepts the EU standard contractual clauses for transfers from Switzerland, adapted to Swiss law: among other points, the FDPIC as supervisory authority, the law governing contractual claims, and references to the GDPR read as references to the FADP. Source: FDPIC, transfer of personal data on the basis of standard data protection clauses (https://www.edoeb.admin.ch/dam/en/sd-web/smvG75WY5Vsi/%C3%9Cbermittlung%20von%20Personendaten%20-%20SCC_EN.pdf). ## The three regimes side by side | EU | UK | Switzerland Adequacy | Commission decisions (Art. 45 GDPR) | UK adequacy regulations, "not materially lower" test | Federal Council list, Annex 1 to the Data Protection Ordinance United States | EU-US Data Privacy Framework (participating organisations) | UK Extension (participating organisations that joined it) | Swiss-US Data Privacy Framework (certified companies) Contract tool | EU standard contractual clauses | IDTA, or UK Addendum to the EU clauses | EU clauses with Swiss adaptations Assessment | Transfer impact assessment (Clause 14) | Transfer risk assessment | See the FDPIC's guidance ## Checklist - For each transfer, identify which regime applies (EU, UK, Swiss, or several) - For US recipients, check the official list for the EU Framework, the UK Extension and the Swiss Framework separately - UK: IDTA or UK Addendum in every contract that relies on safeguards - Switzerland: EU clauses with the Swiss adaptations, or a listed country - Risk assessments documented for each contract-based transfer - Review dates set, including after any ruling on the EU-US Framework ## Questions people ask Can one contract cover EU, UK and Swiss transfers? Yes. Many groups use the EU clauses, with the UK Addendum and the Swiss adaptations attached. Is a US company in the EU Framework automatically covered for UK data? No. It must also have joined the UK Extension. Switzerland has its own framework and certification. Did the 2025 UK Act make transfers easier? It changed the test to "not materially lower" protection. How far that changes outcomes in practice will depend on the government's decisions and the ICO's guidance. Does the EU still treat the UK as adequate? Yes, under the decisions renewed in December 2025. ## Update log - 16 October 2026: first published; sources checked and citations added the same day. ### Data Clean Rooms URL: https://northend.law/guides/data-clean-rooms Section: Privacy-enhancing technologies By Wenceslao M. Spiegel. Published 2026-01-28; last reviewed 2026-01-28. General information, not legal advice. Data Clean Rooms (DCRs) promise privacy-preserving collaboration, but the FTC has warned that marketing claims often exceed reality. Understanding what DCRs can and cannot do is essential for compliant data partnerships. Key points: - Understand DCR privacy claims vs. reality - Configure access controls and query limits - Ensure GDPR/CCPA compliance in data sharing - Address FTC concerns about DCR misuse #### What DCRs actually provide A Data Clean Room is a secure environment where multiple parties can analyze combined datasets without directly sharing raw data. DCRs use technical controls, query restrictions, differential privacy, aggregation thresholds, to limit what each party can learn. However, the FTC's November 2024 guidance emphasizes that DCRs are tools, not magic shields. Misconfigured DCRs can still enable re-identification or leak sensitive insights. #### FTC concerns and enforcement risks The FTC has specifically warned against overstating DCR privacy benefits. Key concerns include: insufficient anonymization leading to re-identification, query patterns that reveal individual-level data, and marketing claims that misrepresent privacy protections. The Commission views DCR privacy promises as potentially deceptive if not technically substantiated. Document your DCR's actual technical controls and limitations. #### GDPR and CCPA compliance Under GDPR, DCR data sharing typically requires a lawful basis, often legitimate interest with proper balancing tests. Data shared into DCRs remains personal data if it can be linked back to individuals. CCPA treats DCR sharing as potential 'sale' or 'sharing' unless exemptions apply. Ensure your DCR agreements address data minimization, purpose limitation, and individual rights obligations across all parties. #### Contractual best practices DCR agreements should specify: permitted query types and output thresholds, audit rights and compliance verification, liability allocation for privacy breaches, data retention and deletion procedures, and incident response obligations. Conduct due diligence on DCR vendors' technical implementations, don't rely solely on marketing materials. ### Identity Solutions (AdTech) URL: https://northend.law/guides/identity-solutions Section: Privacy-enhancing technologies By Wenceslao M. Spiegel. Published 2026-01-28; last reviewed 2026-01-28. General information, not legal advice. As third-party cookies deprecate, identity solutions like UID2, ID5, and RampID have emerged as alternatives. These solutions carry their own privacy requirements and regulatory scrutiny. Key points: - Evaluate UID2, ID5, and alternatives - Ensure valid consent for ID creation - Navigate cross-device tracking rules - Prepare for evolving browser restrictions #### Identity solution landscape Alternative identity solutions fall into categories: deterministic (based on authenticated data like email), probabilistic (statistical matching across signals), contextual (not user-based), and hybrid approaches. Each has different accuracy, scale, privacy, and compliance characteristics. No solution perfectly replaces third-party cookies, you'll likely need a portfolio approach. #### Consent requirements for ID creation Creating identity tokens from user data requires valid consent or other lawful basis. Under GDPR, identity solutions for advertising typically require consent. The consent must be specific to the identity solution, blanket 'personalization' consent may not suffice. Ensure your consent flows explicitly mention identity resolution and cross-site tracking purposes. #### Cross-device and cross-site tracking Identity solutions enabling cross-device tracking face heightened scrutiny. Regulators view such tracking as particularly privacy-invasive. GDPR requires transparency about cross-device linking. CCPA categorizes cross-context behavioral advertising as 'sharing' requiring opt-out mechanisms. The UK ICO has specifically examined adtech identity solutions. #### Future-proofing identity strategy Browser restrictions continue evolving, Privacy Sandbox, Safari ITP, Firefox ETP. Regulatory frameworks increasingly target identity solutions. Build flexibility: invest in first-party data strategies, test multiple identity providers, implement server-side approaches, and monitor regulatory developments. Don't bet everything on a single solution. ### Customer Data Platforms URL: https://northend.law/guides/customer-data-platforms Section: Privacy-enhancing technologies By Wenceslao M. Spiegel. Published 2026-01-28; last reviewed 2026-01-28. General information, not legal advice. Customer Data Platforms unify data across touchpoints, enabling personalization but creating significant privacy obligations. CDPs are both a compliance tool and a compliance risk depending on implementation. Key points: - Establish lawful basis for data unification - Manage consent across data sources - Enable privacy rights fulfillment - Vet CDP vendors for compliance #### CDP role in privacy compliance A well-configured CDP can support privacy compliance by: centralizing consent management across channels, enabling efficient data subject request fulfillment, maintaining data lineage and purpose documentation, and enforcing retention policies. However, the same data unification that enables personalization can amplify privacy risks through expanded profiling capabilities. #### Lawful basis for data unification Under GDPR, unifying data from multiple sources requires lawful basis for each use. Consent given for email marketing doesn't automatically extend to combining that data with app behavior for profiling. Document the lawful basis for each data combination and purpose. CCPA similarly requires disclosure of cross-context behavioral advertising and opt-out mechanisms. #### Consent management integration CDPs should integrate with consent management platforms to respect user preferences across all activation channels. Key requirements: sync consent signals in real-time, prevent activation of data where consent was withdrawn, maintain consent records with timestamps and scope, and ensure downstream systems respect consent limitations. Test consent flows thoroughly. #### Vendor due diligence CDP vendors become data processors requiring proper DPAs. Evaluate: where data is stored and processed, subprocessor lists and change notification, security certifications and audit reports, data portability and deletion capabilities, and contractual commitments on data use. Your CDP vendor's compliance posture directly affects your own. ### AI-native law firms: a reference (October 2026) URL: https://northend.law/guides/ai-native-law-firms-reference Section: Guides By Wenceslao M. Spiegel. Published 2026-10-16; last reviewed 2026-10-29. General information, not legal advice. As explained in an article of January 2026 (https://www.linkedin.com/pulse/legal-ai-vs-ai-native-law-firms-sergio-maldonado-zo9te/): “For all the claims of an ‘AI-native law firm’ (a category that is literally exploding), I find it misleading to apply it to a team that simply leverages AI to do the exact same thing: automate a decades-old workflow, stick to the same value proposition. The term should be reserved instead, I believe, for those who assume that everybody else is using AI too - and define their services accordingly. This requires much more serious reinvention, from questioning the logic of document exchanges to challenging previous perceptions of subject-matter expertise… to daring dumping Microsoft Word for good.” With that caveat out of the way, I have proceeded to compare a group of self-declared "AI-native" law firms across multiple criteria, as shared by each specific firm. This guide does not try to persuade anyone about the better suitability of a given firm over another, but rather provides some structure around the way each law firm describes itself. I am quite convinced that, if we stretch the concept a little, every single law firm is in fact an AI law firm, so the differentiation is not likely to last much, but an in-house lawyer can still find certain common traits that may better suit her or his way of allocating work, resources, and knowledge. ## The firms at a glance All descriptions are taken from each firm's website as they stand at the time of our last review. Please feel free to request an update if you find any inaccuracies. Firm and website | How it describes its structure | Where its lawyers practise or are admitted (as stated) | Main work | Pricing published on its site | AI it describes Arceus (arceus.legal (https://arceus.legal/)) | "Arceus is a law firm. Every engagement is handled by our licensed, U.S.-barred attorneys." | United States; states of admission not stated on the pages read | Commercial and employment contracts, and general counsel work, for legal, sales and finance teams and founders; industries named: AI, B2B SaaS, cybersecurity, healthcare, enterprise software, professional services | Contract review $500 per contract (3 to 25 pages); full negotiation $1,000 per contract; drafting $2,000 per document (smaller drafting projects $1,000 to $1,500). States "an 8 hour turnaround or the review is free" | "Legal agents" combined with licensed attorneys and the company's context; a stored memory of each client's context; requests through Slack or email Falcon (falcon.law (https://falcon.law/)) | Legal services by Falcon Law BV, a Belgian company in Brussels; the platform is developed by Falcon AI BV, "a technology company that is not a law firm and does not provide legal services" | Lawyers of Falcon Law BV are admitted in Belgium and members of the Dutch-speaking Brussels Bar; the site lists Ghent/Brussels, London, Paris and Amsterdam | Commercial contracts in Europe: master services, sales and purchasing agreements, NDAs, terms and conditions, DPAs | Monthly plans: €1,000 (up to 6 tasks), €3,000 (up to 20), €5,000 (up to 40); NDAs and template documents free and unlimited; annual plan invoiced monthly; 24-hour turnaround on the first two plans | "Our AI runs the first review, adapted to your context"; a lawyer signs off; documents used only for the client's work, "never to train models for anyone else"; Slack, Teams or email General Legal (general.legal (https://general.legal/)) | General Legal, LLP, Union City, California (site footer); describes itself as "The AI-Native Law Firm for Growing Companies" | "US barred attorneys"; states of admission not stated on the pages read | Growing companies; nine practice areas listed ("9 practices and growing"), including technology transactions, employment, data privacy, financial regulatory, blockchain and cryptocurrency, emerging companies and venture capital, tax, defense tech and government contracts, and AI commercial and infrastructure transactions | Flat fees quoted before work: simple contract review $250 per contract (3 pages or fewer); standard review $500 (3 to 25 pages); full negotiation $1,000; drafting $2,000 per document (smaller drafting projects $1,000 to $1,500); a fuller fixed-fee schedule on the pricing page | "Our AI tools", by which its attorneys are "accelerated"; matters sent through its web application, Slack, email or MCP Kyra Law (kyralaw.co.uk (https://kyralaw.co.uk/)) | Kyra Law Limited, London (company no. 17111866). "Kyra Law itself is not authorised or regulated by the Solicitors Regulation Authority"; its solicitors are regulated by the SRA individually; it does not take reserved legal activities | England and Wales (SRA-registered solicitors) | Commercial contracts, fundraising (including SEIS/EIS), team and equity, and bulk contract review, for high-growth UK companies | Typical ranges plus VAT: simple contracts £500 to £1,000; more advanced contracts £1,000 to £3,000; a funding round £5,000 to £10,000 | Its own platform ("It's our own. We built it specifically to do legal work"), described as a "proprietary AI platform" that "plugs into all of the major frontier models", and a network of agents; every output verified by a solicitor with a practising certificate in England and Wales; Slack, email or WhatsApp North End Law (northend.law), our own firm | North End Law Corporation, a California professional law corporation (number 31414, approved by the State Bar of California), San Francisco; software from an affiliated technology company is included in its plans | Founding attorney has been admitted in California, Spain (Abogado, Ilustre Colegio de la Abogacía de Madrid), and England & Wales (Solicitor) | Focus: privacy and AI governance for technology companies, alongside the client's existing firm; also copyright, agentic contracts, and crypto and digital assets | Privacy Counsel $1,800, $2,450 or $3,575 a quarter; Fractional AI Governance Lead $920 a month; one-off projects at a flat fee quoted in advance | Its own AI stack: self-hosted todo.law and LQ.AI stacks (open-source software running open-weight models) on firm hardware for privileged work; LQ.AI is an open-source project (Apache-2.0) of the LegalQuants community, to which the firm contributes; its own developments include attorney-built Skills and Knowledge Bases (portable to Donna and other legal AI) and a server for AI agents (MCP); a private cloud running open-weight models for confidential work; commercial models only for anonymised questions of law; citation verification with a log Omnilex Partners (omnilex.law (https://omnilex.law/)) | "An independent law firm, built natively on AI"; "Technology and operations by Omnilex AI", which it calls "the technology company built alongside it" | Not stated on the pages read; the site lists partners in San Francisco and Zurich and an external advisor in New York | Funds and transactions, regulation and compliance, contracts and negotiations; services listed include FINMA and financial licensing, crypto, blockchain and MiCA, SECO and special permits, shelf companies, IP and trademarks, and general counsel | No figures. "Every matter is scoped and priced before work begins: a fixed fee or a fee range" | "The AI runs the research, the first analysis and the first draft immediately; the attorney reviews and decides"; works on the Omnilex AI system; Outlook, Slack or its client portal Soxton (soxton.ai (https://soxton.ai/)) | Site footer: "Soxton Legal". Its terms say an attorney-client relationship begins only with an engagement letter and a conflicts check. The pages read do not state which entity provides the legal services. | Not stated on the pages read | Founders and growing businesses: ongoing counsel and ownership support, contracts | Annual membership $1,000 a year ("unlimited 15-min consultations and contracts at 56% off"); pay-as-you-go from $200 per contract or engagement | "AI-accelerated drafting"; "Our AI applies relevant legal precedent and attorney-built workflows"; review by a lawyer; 24-hour turnaround stated; the site links to an MCP page (not read) Vector Legal (vectorlegal.com (https://vectorlegal.com/)) | Legal services by Vector Legal, P.C., "a licensed California law firm"; the VectorOS platform is built by Vector Systems, PBC | "Licensed in California, New York, and Washington" (its llms.txt) | Startups from formation through Series B and beyond, growth and public companies, venture funds: formation, financing, commercial contracts, governance, ongoing counsel, M&A. States that it does not take litigation, criminal defense, family law, immigration, personal injury or patent prosecution | No figures. "Most work is billed hourly, at about half the cost of Big Law"; outcome-based pricing "where it fits"; monthly fractional general counsel | "Proprietary software" (its llms.txt): VectorOS, built by Vector Systems, PBC, with an AI agent that "reads the company's own record, cites its sources, drafts real documents, and waits for approval before it acts"; integrations include an MCP server; publishes llms.txt and Markdown versions of its pages ## Profiles Arceus presents itself as a US law firm for commercial contracts, built around fast turnaround and fixed prices. It publishes per-contract prices and a turnaround guarantee, and describes legal agents that gather a client's context and pass work to its attorneys. Its attorneys page does not list states of admission. Falcon is a Belgian law firm for commercial contracts across Europe, with a separate company that builds its software. It publishes monthly plans by task volume and a 24-hour turnaround, and states its bar membership, insurance and complaints procedure on a dedicated page. General Legal is a California limited liability partnership serving growing companies across nine practice areas, with flat fees quoted before work starts. It publishes many comparison guides and free templates. Kyra Law is a London company whose solicitors are individually regulated in England and Wales; it states that the company itself is not regulated by the SRA and does not do reserved work. It publishes fixed-fee ranges and describes its own AI platform. North End Law (our own firm) is a California professional law corporation whose stated focus is privacy and AI governance for technology companies, working alongside the client's existing firm. It describes its own AI stack: self-hosted todo.law and LQ.AI stacks on firm hardware, built on the open-source LQ.AI project and its own developments (Skills and Knowledge Bases, a server for AI agents, and software from an affiliated technology company included in its plans), under a published three-tier policy with citation verification. It publishes quarterly and monthly prices. Omnilex Partners describes itself as an independent law firm, with partners in San Francisco and Zurich and an external advisor in New York, working on the system of a technology company built alongside it. It prices each matter in advance with a fixed fee or a fee range, publishes no figures, and the pages read do not state where its lawyers are admitted. Soxton serves founders with a membership or pay-as-you-go model and published prices. The pages read do not state which entity provides its legal services or where its lawyers are admitted. Vector Legal is a California professional corporation next to a public benefit company that builds its platform. Unlike most firms here, it bills most work by the hour. ## How to read this table - Structure. Several firms sit next to a separate technology company. Ask which entity you contract with, which one holds your documents, and whether the software company sees them. - Admissions. A firm can advise only where its lawyers are admitted, or through local counsel. Check the bar's public register. - Prices. Published prices come with conditions (page counts, task weights, VAT). Compare the conditions, not only the figures. - AI. Every firm here says a lawyer reviews the AI's work. Ask where your documents are processed and whether they are used to train models. ## Questions people ask What is an AI-native law firm? A firm that describes its way of working as built around AI. There is no legal definition. Are these firms regulated? Each firm's lawyers are regulated where they are admitted. Several firms state that a separate technology company builds their software, and one states that the company itself is not regulated by its bar regulator while its solicitors are. Are AI-native law firms real law firms? Some are, and some are software companies that work with outside lawyers. The test is simple: a law firm is owned and run under the rules of a bar, and a named lawyer admitted in your jurisdiction is responsible for the advice. The table above shows, for each firm, how it describes its structure and where it states its lawyers practise or are admitted. Which firm should we choose? This guide does not say, and no firm here, including ours, can be compared on a single scale. Compare the work you need, the jurisdictions, and the conditions of each price. ### Trademarks URL: https://northend.law/copyright/trademarks Section: Copyright By Wenceslao M. Spiegel. Published 2026-01-23; last reviewed 2026-01-23. General information, not legal advice. Your trademark is often your most valuable asset. It's not just a logo, it's your brand identity, customer trust, and market position wrapped into legal protection. Here's what every founder needs to know. Key points: - Search USPTO before finalizing your brand name - File trademark applications in key markets early - Monitor for infringement and enforce your rights - Register domain variants to prevent cybersquatting #### Search before you commit Before finalizing your company or product name, conduct a comprehensive trademark search. Start with the USPTO database (free), but also check state registrations, common law uses, and domain availability. A conflicting mark can force a costly rebrand later, or worse, litigation. #### File early, file strategically Trademark rights in the US come from use, but registration provides crucial advantages: nationwide priority, presumption of validity, and access to federal courts. File your application as soon as you're using the mark in commerce. Consider international filings (Madrid Protocol) if you plan to expand globally. #### Choose strong marks Trademarks exist on a spectrum from weak to strong: generic (unprotectable) → descriptive (weak) → suggestive → arbitrary → fanciful (strongest). 'Computer Store' for a tech shop is weak; 'Apple' for computers is arbitrary and strong. Invest in a distinctive mark. #### Protect and enforce consistently Once registered, actively monitor for infringement. Set up Google Alerts, use trademark watch services, and review new trademark applications in your space. Address infringement promptly, ignoring violations can weaken your rights over time. ### Copyright URL: https://northend.law/copyright/copyright-basics Section: Copyright By Wenceslao M. Spiegel. Published 2026-01-23; last reviewed 2026-01-23. General information, not legal advice. As a tech company, your code, content, and creative works are core business assets. Understanding copyright ensures you own what you create and respect others' rights. Key points: - Ensure work-for-hire agreements with contractors - Document code ownership and licensing clearly - Register key works for enhanced legal protection - Respect open source licenses in your stack #### Ownership starts with creation Copyright protection arises automatically when original work is fixed in tangible form, no registration required for protection. However, you need clear agreements about who owns what. Without a work-for-hire agreement, independent contractors may own copyright in code they write for you. #### Work-for-hire fundamentals For employees, work created within the scope of employment is automatically owned by the employer. For contractors, you need an explicit written agreement assigning copyright. Use clear language: 'Contractor hereby assigns to Company all right, title, and interest in any work product created under this Agreement.' #### Open source compliance matters Your software likely includes open source components. Different licenses have different obligations: MIT and Apache are permissive; GPL requires sharing modifications. Maintain a software bill of materials (SBOM) and ensure your use complies with each license's terms. #### Registration unlocks remedies While optional, copyright registration before infringement (or within 3 months of publication) enables statutory damages and attorney's fees in litigation. For key works, your core product code, marketing materials, proprietary content, registration is worth the modest cost.