THE REGULATORY CLOCK
The floor is already law
AI accountability regulation is in active enforcement, not on the horizon. This page tracks the laws that demand evidence records and named human accountability for AI outputs — classified the way a Thread classifies claims: green only when we link the official primary source, yellow when enacted, gray while in motion. Not legal advice.
NEXT DEADLINE
— DAYS
—
5 IN FORCE · 2 ENACTED · 3 IN MOTION
In force
ENFORCEABLE NOW · GREEN ONLY WITH A PRIMARY SOURCE LINKED
EU AI Act — Regulation (EU) 2024/1689IN FORCE
LIVE NOW — ART. 50 TRANSPARENCYAs of Aug 2, 2026, chatbots must tell users they are AI, and AI-generated or manipulated media must be labelled and carry machine-readable provenance marks.
DEFERRED — HIGH-RISK REGIMEThe Digital Omnibus on AI moved the high-risk obligations — technical documentation, record-keeping, and human oversight, with penalties scaled to global revenue — to Dec 2, 2027 for stand-alone (Annex III) systems, and Aug 2, 2028 for AI embedded in regulated products (Annex I).
THE ARTIFACT IT DEMANDSA per-output evidence record plus a named human sign-off: the Thread and the Human Validation Gate, purpose-built. The demand didn’t disappear — it got a later date.
OFFICIAL TEXT — EUR-LEX ↗ · EC — AUG 2 TRANSPARENCY ENFORCEMENT ↗ · EU PARLIAMENT — DIGITAL OMNIBUS ON AI ↗
Colorado AI Act — SB 24-205IN FORCE
WHAT IT REQUIRESA duty of reasonable care for high-risk AI in consequential decisions — employment, lending, housing, insurance — with impact assessments and adverse-decision notices.
THE ARTIFACT IT DEMANDSA defensible per-decision record of what the system actually relied on — an evidence chain you can hand to the Attorney General.
NYC Local Law 144 — Automated Employment Decision ToolsIN FORCE
WHAT IT REQUIRESIndependent bias audits before automated hiring tools are used, published audit summaries, and notice to candidates.
THE ARTIFACT IT DEMANDSAudit-ready, per-decision documentation of automated employment decisions — proof of what the tool did, not a claim that it behaved.
SEC Regulation AC — Analyst CertificationIN FORCE
WHAT IT REQUIRESResearch reports must carry a certification from the named analyst that the views expressed are genuinely theirs.
THE ARTIFACT IT DEMANDSProof that named-human certification is already how regulated industries work — two decades before AI. The Human Validation Gate is that mechanism, generalized to AI outputs.
FSC v. Paxton — age verification upheldIN FORCE
WHAT IT ESTABLISHEDState age-verification requirements for restricted content are constitutional. The age-gating wave across US states now has a settled legal foundation.
THE ARTIFACT IT DEMANDSAge gates don't disappear when the visitor is an agent acting for a human. Platforms will need a verifiable record of the verified person behind automated traffic — sooner than anyone planned for.
Enacted, clock running
SIGNED INTO LAW · EFFECTIVE DATES SET OR CODIFIED SOURCING PENDING
California AB 1043 — Digital Age AssuranceENACTED
WHAT IT REQUIRESAge assurance moves into the operating system: OS providers must collect age at account setup and signal age brackets to every app.
THE ARTIFACT IT DEMANDSEvery app consuming those signals needs a record of what it received and what it relied on — per interaction, inspectable after the fact.
New York Safe By Design — FY27 budgetENACTED
WHAT IT REQUIRESHighest-protection default parental controls on major platforms and restrictions on AI chatbot features for minors.
THE ARTIFACT IT DEMANDSCompliance means showing which protections applied to which interaction — a per-interaction record, not a policy PDF.
CODIFIED TEXT SOURCING PENDING — THIS ENTRY TURNS GREEN WHEN WE PIN THE PRIMARY SOURCE
In motion
DIRECTION OF TRAVEL · STRONG SIGNAL, NOT YET SETTLED LAW
New York S9051B — companion chatbots and minorsIN MOTION
WHAT IT WOULD REQUIREProhibits companion chatbots for minors and restricts systems from presenting as humans or encouraging harmful behavior.
THE DIRECTIONAI systems interacting with protected classes of users will need to prove, per interaction, what they were and what they did.
Supervisory convergence — financial and cross-sector regulatorsIN MOTION
THE PATTERNCapital-markets supervisory guidance and national AI governance frameworks keep converging on the same two requirements: lifecycle documentation of AI systems, and accountability that terminates in an identifiable human — not in a score.
THE DIRECTIONAutomated risk surfacing is treated as necessary but not sufficient. The missing piece is always the named person on the record.
Agent identity frameworksIN MOTION
THE PATTERNMultiple jurisdictions are moving toward registered, verifiable identities for AI agents as agent traffic grows toward parity with human traffic.
THE DIRECTIONWhen agents carry identities, their outputs need provenance attached to those identities. Identity says who acted; provenance says what the action was based on. Regulators will want both.
LAST REVIEWED 2026-08-03 · THIS PAGE CLASSIFIES LAWS THE WAY THREADS CLASSIFY CLAIMS — GREEN ONLY WITH A PRIMARY SOURCE, AND ENTRIES CHANGE ONLY WHEN THE LAW DOES · THIS IS A CATEGORY MAP, NOT LEGAL ADVICE — CONSULT COUNSEL FOR YOUR OBLIGATIONS