AI Compliance: Navigating Shifting Legislation and Core Principles

"Are we compliant with AI regulation?" There's no longer a single answer to that question in the US — there are several, and they don't agree. While everyone watched Washington, the states moved. A company operating nationally now faces a patchwork: rules in California, Texas, Illinois, Utah, and New York City that are live now, with more arriving on staggered dates. The problem isn't any one law. It's that "compliant" is no longer a status — it's a matrix of where you operate, who your AI affects, and what each jurisdiction demands. A federal preemption fight overhead may eventually simplify it. It hasn't yet. For most organizations the honest answer to "are we compliant?" today is a question: "compliant where, with what, affecting whom?" If you can't answer that, the first gap isn't legal. It's knowing what AI you're running at all.

Does legal AI have a shelf life? The law changes constantly, but usually slower than the world around it, and slower than technology itself. That gap is actually the interesting part. With the right data engineering behind it, AI can start making predictive reads on where the law is heading, sometimes faster than the technology tracking it can even keep pace with itself. So no, legal AI does not have a shelf life. If anything, the gap between how fast the law moves and how fast everything else moves is exactly what makes this space worth building in long term.

AI/Legal Systems/Us Humans Part 8/8 Where this leaves the people who can actually act on it. Legislators: the disclosure obligation from Part 5 is the fastest-moving priority — least new infrastructure required, quickest reduction in the harm described in Part 4. Regulators: machine-readable rulemaking can be piloted under existing rulemaking authority in most jurisdictions, without waiting for new legislative grant. Judiciaries: evidentiary standards for AI-derived capability differentials can develop within existing equality-of-arms jurisprudence now, rather than waiting for a legislative definition that will likely lag the harm by a full technology cycle. Legal training: curricular weight shifts toward disclosure literacy and algorithmic-advantage auditing — the competencies the formalized system will require regardless of how quickly formal adoption moves. The real question was never whether AI should be let into law. It's already inside legal outcomes, through every sophisticated actor using it without disclosure. The only open question is whether that stays concealed — or becomes visible and governed, the way this body of law has handled every comparable problem on record. Full long-form version available on request. (8/8 — end)

The U.S. wants less AI regulation. The EU wants more structure. Your compliance program can't wait for them to agree. At the recent G20 Innovation Ministerial, Washington doubled down on a deregulatory approach to AI, while Brussels continues building out the binding, risk-tiered EU AI Act framework. For companies operating on both sides of the Atlantic, that's not a policy footnote. It's a compliance problem, today. In our latest Insights piece, we break down what this regulatory divergence means for: → Contract and vendor risk allocation → M&A diligence scope → Board-level AI governance → Where to focus your compliance efforts now, not after the gap closes Read the full analysis on our site. 🔗 https://lnkd.in/g9DVztiW #AICompliance #CorporateLaw #EURegulation #AIGovernance #HolonLaw

Europe just flipped the switch on the world’s most ambitious AI law, and U.S. technology companies aren’t exempt. With Article 50 transparency rules now live and general‑purpose AI oversight fully enforceable, the question isn’t if the Act applies — it’s where your AI touches the EU.

AI/Legal Systems/Us Humans Part 5/8 What formalizing AI into law actually requires. Five elements, not a slogan — four here, one in the next post. Machine-readable rulemaking, issued by the rule-making body itself alongside the prose version — not reverse-engineered downstream by whichever party can afford to build that translation privately. A disclosure duty for AI-derived counterparty advantage, categorized within existing information-asymmetry doctrine rather than requiring an entirely new legal framework. This doesn't prohibit the practice. It makes it visible enough to weigh. Liability rules that name the capability gap explicitly — allocation defined in advance between model provider, deploying party, and disclosure holder, rather than litigated from first principles each time. Institutional AI parity for courts, regulators, and legal aid structures, funded as a precondition for the first three elements to remain enforceable — not as a discretionary efficiency initiative. None of this is speculative technology. Every element is buildable today. What's missing is institutional priority. A fifth element — how to verify output at machine scale without checking every output by hand — needs its own post. That's next. (5/8)

I have a new article out in The European Business Review on responsible AI and where regulation is heading. 🤖 Here are a few of the takeaways: ➡️ You can outsource labour to a machine. You cannot outsource the consequences. ➡️ The Digital Omnibus deferred the EU's high-risk obligations to December 2027, but transparency obligations landed on schedule. Brussels is adjusting sequencing, not softening intent. ➡️ US federal legislation looks unlikely for now, while states keep passing AI laws anyway. I think the current federal trough is a phase, not an equilibrium. ➡️ For multinationals: find the common denominators across AI regimes, make those your global baseline, and layer jurisdiction-specific requirements on top. ➡️ The companies that win will be the ones that can prove human control, not merely assert it. Full article here: https://lnkd.in/ebtyPReX #ResponsibleAI #AIGovernance #EUAIAct #AIRegulation #GRC #Compliance

Article 50 of the EU AI Act came into force on 2 August 2026. Since then legal teams have treated all obligations under Article 50 as applicable now. It's more nuanced than that. Article 50 contains two distinct obligations. The first: ensuring users know they are interacting with an AI system. That is applicable now. The second: marking AI-generated content in a machine-readable format. That is not yet fully live for everyone. Read the Omnibus alongside Article 50. Providers who already had systems on the market before 2 August 2026 have until 2 December 2026 to comply with Article 50(2). That grace period exists specifically for providers already in the market. Anthropic, for example, has not yet fully complied - and that is entirely consistent with the regulation as written. Article 50 alone doesn't make this obvious. The Omnibus is doing critical work here and you need to review both documents to get the timeline right. Legal teams - this is where the real work starts. Do you have a technical solution for Article 50(2) in place before December? And if you're a deployer, do you actually know where your AI providers stand? I have some ideas... Watch this space. #EUAIAct #AIGovernance #Article50 #AICompliance #LegalTech #InHouseLegal #AIRegulation

🔥 It's OFFICIAL: A new era for Artificial Intelligence has begun! 🔥 The EU AI Act, the world's FIRST comprehensive law regulating AI, officially comes into force today, May 21st. 🇪🇺 This landmark legislation isn't just about rules; it's a global statement on building responsible, safe, and ethical AI. What does this mean for YOU? 🤔 Whether you're an AI developer, a tech leader, or a business leveraging AI, expect a robust, risk-based framework. From high-risk system transparency to data governance and human oversight, the goal is clear: foster innovation while safeguarding fundamental rights. Non-compliance comes with hefty fines! 💰 This isn't just Europe's law; it's a global benchmark shaping the future of AI everywhere. How do you foresee this act impacting your projects or industry? Let's discuss! 👇 #EUAIAct #AIregulation #ArtificialIntelligence #TechLaw #Innovation #Compliance #EthicalAI #FutureofTech #DigitalEthics

Everyone celebrated the EU AI Act delay back in May. Then August 2, 2026 arrived, and something did become enforceable that day. I still see people treating "EU AI Act delayed" as the end of the story. It wasn't. Here's the full picture now that the date everyone thought got pushed off has actually come and gone. 📅 What got delayed The EU's Digital Omnibus on AI, now Regulation (EU) 2026/1744, entered into force on July 27, 2026. It pushed the compliance deadline for standalone high-risk systems, the Annex III category, from August 2, 2026 to December 2, 2027. That's real, it's law, and if you're building or buying hiring tools, credit scoring systems, or anything else that falls under Annex III, you genuinely have more runway than the original timeline gave you. ⚠️ What became enforceable anyway Article 50's transparency obligations took effect exactly as scheduled on August 2, 2026: ↳ Disclosing when someone's interacting with an AI system ↳ Labeling AI-generated content ↳ Flagging deepfakes ↳ Disclosing emption recognition and biometric categorization system use Those apply based on what the system does, not which risk tier it sits in, so the delay never touched them. One additional wrinkle worth knowing: the watermarking piece specifically has its own grace period running to December 2, 2026, so that sub-requirement is still ahead of you even though the rest of Article 50 is already live. 🎯 Why the delay still doesn't buy the time people think it does The Annex III delay exists because regulators, standards bodies, and enterprises all admitted the same thing: nobody was ready. That's not a reason to relax, it's confirmation the operational work was already behind schedule before the deadline moved. And now that August 2 has actually passed with Article 50 live, the gap between "we have until 2027" and "some of this already applies to us today" is no longer theoretical. If your team read the May headlines and stood down governance work, that's worth revisiting now, not in 2027. 🔔 Follow Terry Dwelle for practical guidance built for business leaders