Updated April 2026 State-level consumer AI legislation Non-federal scope

What this tracker covers

State-level consumer AI legislation in the United States, captured monthly with out-of-cycle refreshes for enactments, vetoes, and special sessions. The snapshot date is April 2026.

Included: any bill — introduced, pending, enacted, or vetoed — that touches one or more of the six rights covered in the next section (Disclosure, Consent, Mental health, Likeness, Elections, Enforcement).

Excluded: federal proposals, non-US jurisdictions, bills that don't touch the six rights, and bills we haven't yet triaged.

How we classify bills

Every record in the tracker carries two layers of classification: the rights the bill touches (what consumer protection the state chose to require) and the status of the bill (where it sits in the state process).

The six rights

I · DisclosureConsumers are told when AI is making decisions that affect them.
II · ConsentConsumers must opt in before AI is used on them in specific contexts.
III · Mental healthAI used in mental-health contexts faces extra duties around disclosure, referral, and crisis handling.
IV · LikenessConsumers control uses of their voice, face, and performance — including deepfakes.
V · ElectionsAI-generated political content must be labelled; impersonation of candidates is prohibited.
VI · EnforcementConsumers can pursue violations through state AGs or private right of action.

The status ladder

Four statuses reflect what each state chose:

The frontier-governance tag

A bill may additionally carry a frontier-governance tag. This marks bills that impose duties on developers of frontier-capability AI systems — a narrower category than consumer-facing AI. The next section explains the four-criteria threshold that defines the tag.

Frontier governance — the §4.2 threshold

A state bill qualifies as "frontier governance" if it imposes at least one of the following on developers of frontier-capability AI systems:

Pre-deployment evaluations Capability evaluations are required before a qualifying model is released.
Third-party audits An independent party must audit safety claims, not only the developer's internal team.
Compute or capability thresholds Duties scale with training compute, benchmark performance, or other capability signals — not merely company size.
Incident reporting Frontier developers must report qualifying incidents to the state or a designated agency.

These four criteria were chosen because they are the governance levers most specific to frontier-capability systems — the ones that distinguish a "frontier AI bill" from broader consumer-protection legislation. Each state decides independently which combination to require.

How we score states

The scorecard ranks states by a composite score that rewards enacted law, counts active pending bills, and adjusts for how broadly a state has chosen to cover the six rights. The formula:

score = (enacted × 3) + pending − vetoed, adjusted for recency and rights breadth
EnactedWeight 3. Laws on the books carry the most weight because they represent actually-imposed duties.
PendingWeight 1. Bills that have advanced past first committee count as active movement.
IntroducedExcluded from the base score. Tracked in the ledger for completeness.
Breadth bonusThe scorecard's All view adds weight when a state has chosen to cover more of the six rights.

Recency is applied as a mild decay: enactments from the current session and one year back count fully; older enactments lose a small amount of weight per year. This prevents a state that enacted one law in 2019 and nothing since from ranking above a state actively legislating today.

The scorecard also carries a rights-coverage matrix that shows which of the six rights each top-ranked state has addressed. That matrix is sorted by breadth (number of rights covered), not by composite score — it's a companion view, not a ranking.

Data sources & refresh cadence

Dataset Bill records, state scorecards, and citations live in js/asai-data.js — the single source of truth for every surface on this tracker.
Public sources State legislative websites (linked per bill in the ledger), enacted-law citations, and governor press releases. Every external citation is footnoted in the matrix.
Refresh cadence Monthly sweep, plus an out-of-cycle refresh whenever an enactment, veto, or special session changes the picture in a covered state.
Citation integrity Every external URL is verified via scripts/check-citations.js before each ship; the gate must pass at zero failures before a release.

Spotted a stale record or want to flag a bill we've missed? Email press@secureainow.org with the bill ID from the ledger row.