April 2, 2026

Today's briefing is heavily weighted toward U.S. state-level developments; EU coverage is limited to a single civil society publication. The dominant theme is AI governance infrastructure — states are moving simultaneously on third-party verification frameworks, training data transparency, and sector-specific AI rules, while several bills that had been tracking toward passage have stalled or failed outright.

Virginia's dual AI verification bills, covered yesterday as cleared by the legislature, have now produced final vote tallies worth noting. SB384 passed the House 99-0; HB797 passed with an 84-14 margin on a Senate substitute. The near-unanimous vote on SB384 in particular suggests broad political durability for the verification organization model. Both bills now await the Governor. Virginia is emerging as a reference jurisdiction for third-party audit infrastructure, and compliance teams advising AI developers on audit readiness should begin mapping what an independent verification engagement might require under that framework. Watch level: PREPARE (AI developers, enterprise risk and compliance teams, third-party auditors)

Washington State's HB1170, requiring disclosure when content has been developed or modified by AI, has been delivered to the Governor for signature or veto. This is a consumer-facing transparency obligation — affirmative notification, not just internal documentation. Washington's HB2157, the broader high-risk AI bill, has simultaneously been placed in the House Rules 'X' file and will not advance this session. The divergence is instructive: disclosure mandates are clearing where broader liability-based frameworks are not. Digital content platforms, marketing technology operators, and AI-assisted publishing tools should treat HB1170 as likely to be signed. Watch level: PREPARE (digital content platforms, AI content generation tools, marketing technology operators)

New York has two companion bills — S06955 in the Senate and A06578 in the Assembly — both titled the Artificial Intelligence Training Data Transparency Act, both now at third reading. The parallel advancement suggests coordinated bicameral momentum toward requiring generative AI developers to publicly disclose training dataset information on their websites. New York has not enacted these yet, but the procedural posture warrants scoping now. Developers of generative AI products distributed in New York should assess what training data documentation they currently maintain and whether it would satisfy a public disclosure standard. Watch level: PREPARE (generative AI developers, foundation model providers, AI product counsel)

Colorado's HB1139 passed its House Second Reading with amendments — this bill, covered yesterday, has now cleared a material procedural hurdle and moves to the Senate. Health care organizations operating in Colorado should track the Senate amendment text closely, as floor amendments at this stage can materially shift scope. Separately, Colorado introduced HB1263 targeting conversational AI service operators, currently in committee; it is early-stage but fits the pattern of Colorado layering sector-specific AI obligations onto its existing high-risk AI framework. Utah's HB0450 and HB0320 are both transmitted to the Governor, and Wyoming's HB0102 targeting AI-generated deepfakes of minors is now enacted. Wyoming's enactment is the only new law in today's briefing; it is operative now. Watch level: ACT NOW (AI image-generation platforms, content hosting services operating in Wyoming — review HB0102 definitions and liability provisions immediately)

The CDT Europe's fifth EU AI Act brief on general-purpose AI model obligations is civil society analysis, not regulatory action. However, civil society interpretations of GPAI provisions have historically influenced how national competent authorities develop enforcement priorities under the Act. GPAI providers monitoring the EU compliance environment should add this to their regulatory intelligence stack, particularly on systemic risk thresholds. Vermont's H0814 on neurological rights and Maryland's HB1261 on AI-integrated children's toys both remain early-stage, but together they reinforce a visible trend: state legislatures are moving toward product- and domain-specific AI rules rather than waiting for horizontal frameworks to resolve. Watch level: MONITOR (GPAI model providers for CDT brief; neurotech developers and AI toy manufacturers for state-level vertical regulation trends)