50-State Survey: AI Regulation

Which of the 50 states have already passed AI laws, and what is coming next. From deepfakes to high-risk AI rules, this 50-state survey is your source for the details.

By the team at Gavel

Published:

Sep 19, 2025

| Last updated: Feb 3, 2026

Cut drafting time by 90%

Easy intake and document automation to auto-populate your templates.

1) State laws and regulations already in effect

Arkansas: AI-generated content ownership law

• What it does: Clarifies who owns content created by generative AI and certain model-training outputs, with rules around employer ownership of work-related outputs.
• Who it applies to: Creators, employers, and companies working with generative AI in Arkansas.
• Status: Enacted April 2025.

California: Performer likeness laws (AB 2602 and AB 1836)

• What they do: Protect performers, living or deceased, from unauthorized AI-generated digital replicas of their voice or likeness.
• Who they apply to: Anyone creating or distributing AI-generated replicas of performers.
• Status: Signed September 2024.

California: CCPA rulemaking on Automated Decision-Making (ADMT)

• What it does: In July 2025, the CPPA adopted rules on automated decision-making, risk assessments, and cybersecurity audits. They take effect once approved by the Office of Administrative Law.
• Who it applies to: Businesses subject to the CCPA using ADMT.
• Status: Adopted; awaiting final approval. If approved on the first submission, they could be operative by December 1, 2025.

Colorado: Colorado Artificial Intelligence Act (SB 24-205)

• What it does: Creates the first broad state law for “high-risk” AI, things like hiring, housing, credit, health, and education decisions. It sets obligations for both developers and deployers (risk controls, disclosures, incident reporting) and gives a safe harbor for those following recognized frameworks.
• Who it applies to: AI developers and deployers in Colorado.
• Effective: February 1, 2026.

Montana: Right to Compute Act (2025)

• What it does: Protects lawful access to computing resources and requires risk management policies and shutdown options for AI that controls critical infrastructure.
• Who it applies to: Operators of AI in critical infrastructure, and residents more generally for computing rights.
• Status: Signed April 2025.

New York: State agency AI transparency law

• What it does: Requires state agencies to keep track of and publicly disclose automated decision tools they use, while adding safeguards for workers.
• Who it applies to: State agencies in New York (not private companies).
• Status: Passed in late 2024; implementation adjusted in 2025.

North Dakota: Anti-harassment by robot/AI (2025)

• What it does: Expands harassment and stalking laws to cover AI-powered robots and some drones.
• Who it applies to: People using robots or drones to harass others.
• Status: Signed April 23, 2025.

Oregon: “Nurse” title restrictions (2025)

• What it does: Prohibits AI agents or other non-human entities from using protected healthcare titles like “nurse.”
• Who it applies to: AI agents and anyone misrepresenting AI as a healthcare professional.
• Status: Passed March 2025.

Tennessee: ELVIS Act

• What it does: Expands the state’s right of publicity to cover AI-created voice and likeness cloning, giving artists a way to sue for unauthorized use.
• Who it applies to: Anyone using another person’s voice or likeness without consent, including AI companies.
• Effective: July 1, 2024.

Utah: Artificial Intelligence Policy Act (SB 149, 2024) plus 2025 amendments

• What it does: Requires disclosure when consumers are interacting with AI in certain contexts and creates an AI office. Amendments in 2025 added consumer protection rules, narrowed some disclosure requirements, introduced standards for mental health chatbots, and tightened impersonation protections.
• Who it applies to: Businesses and professionals using AI with Utah consumers, including those offering mental health chatbots.
• Effective: May 1, 2024 (core law); May 7, 2025 (amendments).

Election deepfakes: Multiple states

• What they do: Dozens of states have passed laws restricting deceptive AI-generated political content near elections or requiring clear disclosures. Some require technical markers or metadata.
• Who they apply to: Campaigns, political actors, platforms, and distributors of AI-generated political content.
• Status: More than 25 states have passed laws as of mid-2025.

Note: A number of other states have passed narrower measures, like creating task forces, studying AI’s impact, or setting procurement standards. The pace of state-level activity is quickening.

2) What's are states considering next (and when)

California: Frontier AI safety bill (SB 53, 2025)

• What’s proposed: Would require certain developers of large “frontier” models to publish safety plans, disclose pre-deployment testing, and report incidents. It’s narrower than 2024’s vetoed SB 1047.
• Timing: Passed the Legislature in September 2025 and is waiting for the Governor’s decision. California’s deadline to sign or veto bills is in October.

Colorado-style “high-risk AI” bills elsewhere

• What’s proposed: Similar to Colorado’s framework, these bills would place obligations on developers and deployers and add discrimination-prevention requirements. Several 2025 attempts stalled but will likely come back.
• Timing: Most states will take them up again when legislatures reconvene between January and June 2026.

Right-of-likeness/anti-deepfake proposals

• What’s proposed: Expanding protections for voice and image rights, including faster takedowns and new platform duties, tracking ideas in the federal NO FAKES Act.
• Timing: Expected to continue through late 2025 and into the 2026 sessions.

Government use and transparency of AI

• What’s proposed: Following New York’s lead, more states are considering inventories of AI systems used by agencies and restrictions on automated eligibility decisions.
• Timing: Likely to surface in 2025–2026 sessions, often tied to budget planning.

Election and campaign AI rules

• What’s proposed: More states are looking at disclosure or outright bans on deceptive political deepfakes, with some pushing for embedded watermarks or labels.
• Timing: Expect additional laws before the 2026 primaries, with some states possibly acting in late 2025.

This list was last updated September 19, 2025.