State AI Laws by State: The 2026 US Tracker

State AI laws by state are now the main place AI policy is being made in the US. In 2025, state lawmakers introduced 1,208 AI-related bills, the first year every state had introduced at least one. 145 became law.

By March 2026, lawmakers had already introduced 1,561 new AI bills across 45 states. The pace is faster than any prior tech-law category, including data privacy.

This guide breaks down state AI laws by state — what is enacted, what is pending, what each rule covers, and when it takes effect. Use the in-page links to jump to your state.

State AI laws at a glance

Five states stand out in 2026: Colorado, Texas, California, Utah, and Illinois. Each has at least one AI law in force or set to take effect this year.

The other 45 states fall into three buckets: deepfake and election laws, narrow employment or healthcare AI rules, or pending bills that have not yet been enacted. The snapshot below lists the most important rules first.

• Colorado — SB 24-205 (Consumer Protections for AI). Effective June 30, 2026 (delayed from Feb 1, 2026). Federal court enforcement pause as of April 27, 2026.
• Texas — TRAIGA / HB 149. Effective January 1, 2026. Bans intentional harm, social scoring, CSAM deepfakes.
• California — SB 53 (TFAIA, frontier transparency) + AB 2013 (training data). Both effective January 1, 2026.
• Utah — SB 149 (AI Policy Act). Effective May 1, 2024. Generative AI disclosure on request and in high-risk uses.
• Illinois — HB 3773 (AI in employment). Effective January 1, 2026. Amends the Illinois Human Rights Act.
• New York City — Local Law 144 (bias audits for automated employment decision tools). In force since 2023.
• 38+ states — Deepfake / election AI laws (nonconsensual intimate imagery, candidate-impersonation rules).

Colorado — SB 24-205 (Consumer Protections for AI)

Colorado SB 24-205 is the broadest US state AI law. It applies to developers and deployers of "high-risk AI systems" that make consequential decisions about Coloradans.

The law was scheduled to take effect February 1, 2026, but was delayed by SB 25B-004 to June 30, 2026. On April 27, 2026, a federal court paused enforcement while lawmakers consider further amendments.

See our Colorado AI law compliance guide for a step-by-step compliance plan.

• Covers high-risk AI used in employment, housing, lending, insurance, healthcare, education, and government services.
• Developer duties: documentation, impact assessment summary, and risk-management disclosure to deployers.
• Deployer duties: impact assessment, written risk-management policy, consumer notice, right to correct, right to appeal.
• Safe harbor for organizations using NIST AI RMF or similar recognized frameworks.
• Colorado AG has exclusive enforcement authority; no private right of action.
• Most recent status: enforcement paused April 27, 2026; further amendments expected before June 30, 2026 effective date.

Texas — TRAIGA (HB 149)

Texas TRAIGA was signed by Governor Abbott on June 22, 2025 and took effect January 1, 2026. The final version is much narrower than the original draft.

TRAIGA focuses on a short list of banned uses, plus rules for state-government AI. It is not a broad "high-risk AI" regime like Colorado's.

• Bans intentional development or deployment of AI that incites self-harm or criminal activity.
• Bans AI-generated child sexual abuse material and nonconsensual deepfake pornography.
• Bans government social-scoring AI.
• Requires Texas state agencies to disclose AI use to consumers.
• Civil penalties enforced by the Texas Attorney General.
• No "high-risk AI" impact assessment regime — that piece was cut from the final bill.

California — SB 53 and AB 2013

California passed two major AI laws in 2025: SB 53 (Transparency in Frontier Artificial Intelligence Act) and AB 2013 (training-data transparency). Both took effect January 1, 2026.

SB 53 targets the largest AI developers — those training models above 10^26 FLOPs. AB 2013 applies more broadly to any generative AI developer that makes its system available in California.

• SB 53 (TFAIA) — Public, standardized safety disclosures for frontier model developers.
• SB 53 threshold — Training compute above 10^26 FLOPs (including fine-tuning and material modifications).
• SB 53 — Whistleblower protections and mandatory reporting of serious safety incidents.
• AB 2013 — Generative AI developers must post a summary of their training data on their website.
• AB 2013 — Applies to systems released on or after January 1, 2022 and requires re-posting on substantial modification.
• California did not pass SB 1047 in 2024; SB 53 is the narrower replacement that survived the Governor's veto threshold.

Utah — SB 149 (AI Policy Act)

Utah was the first US state to pass a generative AI consumer-protection law. SB 149 took effect May 1, 2024 and has been amended twice since.

The law requires businesses to disclose generative AI use on request, and to disclose proactively in "high-risk interactions" such as health, financial, or legal advice.

• Effective May 1, 2024. Amendments effective May 2025 narrowed the scope to regulated professions.
• Disclose generative AI use when a consumer explicitly asks.
• Mandatory upfront disclosure in high-risk interactions (mental health, healthcare, legal, financial).
• Removes the "the machine did it" defense — businesses are liable for AI-generated statements.
• Civil penalties up to $2,500 per violation; up to $5,000 for repeat violations in court.
• Created the Utah AI Learning Laboratory to test innovation-friendly regulatory approaches.

Illinois — HB 3773 (AI in employment)

Illinois HB 3773 amends the Illinois Human Rights Act to cover AI use in employment decisions. It took effect January 1, 2026.

The law prohibits employers from using AI in a way that has a discriminatory effect on protected classes — and requires notice when AI is used in employment decisions.

• Effective January 1, 2026.
• Amends the Illinois Human Rights Act (IHRA).
• Prohibits AI use in employment decisions with discriminatory effect on protected classes.
• Requires employers to notify employees and applicants when AI is used.
• Enforced by the Illinois Department of Human Rights.
• Stacks with the existing Illinois AI Video Interview Act (in force since 2020).

New York — NYC Local Law 144 and state bills

New York City's Local Law 144 has been in force since July 2023. It requires bias audits and candidate notice for automated employment decision tools (AEDTs).

At the state level, New York has introduced several AI bills in 2025–2026 — including frontier-model rules and election-deepfake laws. As of June 2026, no broad state-level AI Act has been enacted.

• NYC Local Law 144 — Annual bias audit by an independent auditor for any AEDT used in NYC hiring.
• NYC LL 144 — Public summary of audit results and candidate notice required.
• New York election-deepfake law — Restricts AI-generated political ads within 60 days of an election.
• New York state SAFE for Kids Act — Restricts addictive social-media algorithms for minors.
• Multiple frontier-AI bills pending in the 2025–2026 session.

Other states — deepfake, election, and narrow-scope AI laws

Most other states have not yet passed a broad AI law. But the majority have at least one narrow AI rule on the books — usually targeting deepfakes, election ads, or specific sectors like healthcare or insurance.

The list below is not exhaustive. Treat it as a snapshot of the most-cited state rules as of June 2026.

• Tennessee — ELVIS Act (Ensuring Likeness Voice and Image Security). Effective July 1, 2024. AI voice cloning.
• Arizona, Florida, Michigan, Minnesota, Texas, Washington — Election deepfake laws.
• Virginia — HB 2094 (AI Developer and Deployer Act). Vetoed by Governor in 2025; no broad AI law enacted.
• Maryland — Multiple AI study commissions and narrow AI bills in 2025–2026.
• Massachusetts — Pending broad AI bill in the 2025–2026 session; not yet enacted.
• Washington — Multiple AI-in-government bills enacted; broad consumer AI law still pending.
• New Jersey — AI bias-audit bill pending; deepfake law enacted.

How to comply with multiple state AI laws

Most companies operate in more than one state. The good news: the major state AI laws overlap a lot. A single NIST AI RMF program covers most of what Colorado, California, and the EU AI Act require.

Below is the order of operations we use to build multi-state AI compliance programs at Layer3.

• Build a single AI inventory across all states where you operate.
• Adopt NIST AI RMF 1.0 as your baseline — Colorado gives explicit safe-harbor credit.
• Map your AI use cases to "high-risk" categories — Colorado SB 24-205 and the EU AI Act use similar definitions.
• Write a one-page AI use policy that covers disclosure, prohibited uses, and human review.
• Stand up an AI vendor due-diligence checklist that asks for model cards, safety policies, and SOC 2 reports.
• Run impact assessments for any "high-risk" use — covers Colorado, the EU AI Act, and most pending state bills.
• Track new bills with NCSL's AI legislation database and re-check this tracker quarterly.