US AI Safety Map · free

There is no federal answer. There are 12 state answers.

Companion-chatbot safety, frontier-model safety and AI transparency are being legislated state by state, on different dates, with different triggers. This is what is actually on the books today — read from the corpus rather than from a news round-up.

See the map Read the 17 laws

In the corpus

17 laws across 12 states

Every date

The label the statute itself carries

Three layers, fifty states

Most states are blank. That is the finding.

Select a layer to isolate it. A blank tile means no law on that layer today — it does not mean a deployment there is unregulated, because privacy, employment and consumer-protection statutes reach AI without ever using the word.

At least one law on these layers None on these layers yet

A tile is not a determination. This map answers which statutes exist and when they commence. Whether one reaches your deployment turns on what the tool does, who it touches, and whether you meet the entity threshold — five gates, each answered separately. The map is where the question starts, not where it is settled.

The laws

All 17, with the date each one actually bites.

Grouped by state and ordered by commencement. Several do not take effect for another eighteen months, which is the useful part — it is the window in which changing a deployment is still cheap.

Layer

California

4 laws

AI content transparency

AB 2013California statute requiring developers of generative AI to publish high-level summaries of the data used to train their systems.

In ForceJanuary 1, 2026

Companion chatbot safety

Cal. Bus. & Prof. Code div. 8, ch. 22.6, §§ 22601 to 22606 (Companion Chatbots)California statute (Business and Professions Code, division 8, chapter 22.6, Companion Chatbots, added by Stats.

In ForceEffective January 1, 2026

AI content transparency

SB 942California statute requiring providers of publicly accessible generative AI systems to make a free disclosure verification tool available and to include latent provenance disclo…

In ForceAugust 2, 2026 (providers); January 1, 2027 (platforms); January 1, 2028 (capture devices)

Frontier AI safety

Cal. Bus. & Prof. Code Ch. 25.1 (Transparency in Frontier Artificial Intelligence Act)California statute requiring large frontier AI developers to publish frontier AI frameworks, catastrophic-risk assessments, critical safety incident reporting, and whistleblower…

In ForceEffective January 1, 2026

Colorado

1 law

Companion chatbot safety

HB 26-1263Colorado statute requiring an operator of a publicly accessible conversational artificial intelligence service to estimate user age, disclose AI interaction (with enhanced discl…

In ForceAugust 12, 2026 (section codified); operator duties operative January 1, 2027 (§ 6-1-1708(2)-(5)); annual reporting operative July 1, 2027 (§ 6-1-1708(6))

Georgia

1 law

Companion chatbot safety

2026 Ga. Laws Act 518, amending O.C.G.A. tit. 39, ch. 5 (artificial intelligence companion chatbots)Georgia statute regulating conversational AI services — mandatory AI disclosure, minor safety guardrails, privacy and engagement controls, crisis-response protocols, and Attorne…

PendingJuly 1, 2027

Idaho

1 law

Companion chatbot safety

Code §§ 48-2101 to 48-2105 (Conversational AI Safety Act)Idaho Conversational AI Safety Act — operator disclosure when users could believe they interact with a human, heightened minor protections, anti-manipulation rules, crisis-respo…

PendingJuly 1, 2027

Illinois

1 law

Frontier AI safety

SB 315Illinois frontier-AI safety statute requiring large frontier developers to publish catastrophic-risk frameworks, pre-deployment transparency reports, annual independent third-pa…

PendingJanuary 1, 2028 (framework, audits, transparency reports)

Iowa

1 law

Companion chatbot safety

Code ch. 554J (Conversational AI Services; Senate File 2417, 2026 Iowa Acts ch. 1068)Iowa statute (Chapter 554J) regulating conversational AI operators — AI disclosure, minor account-holder safeguards, suicide/self-harm protocols, prohibition on representing the…

PendingJuly 1, 2027

Maine

1 law

AI content transparency

Me. Rev. Stat. tit. 10, § 1500-DD (Required disclosure of use of artificial intelligence chatbot to engage in trade and commerce)Maine statute (10 MRSA §1500-DD) requiring clear and conspicuous disclosure when AI chatbots engage Maine consumers in trade and commerce in a manner that could mislead a reason…

In ForceSeptember 24, 2025

Nebraska

1 law

Companion chatbot safety

LB 525Nebraska session act — §§ 12–18 of Legislative Bill 525, the Conversational Artificial Intelligence Safety Act — operative July 1, 2027, governing conversational artificial inte…

PendingJuly 1, 2027

New York

2 laws

Frontier AI safety

S8828New York's broad AI safety and education statute, addressing developer and deployer obligations for covered AI systems, transparency, and frontier model risk reporting.

PendingJanuary 1, 2027

Companion chatbot safety

York AI Companion LawNew York General Business Law Article 47 — Artificial Intelligence Companion Models — was added by Part U, § 1 of ch.

In ForceNovember 5, 2025

Oregon

1 law

Companion chatbot safety

SB 1546Oregon session act, effective January 1, 2027, governing artificial intelligence companions — systems using artificial intelligence, generative artificial intelligence or emotio…

PendingJanuary 1, 2027

Rhode Island

1 law

Companion chatbot safety

Island S 2195Rhode Island statute (Title 6 Chapter 63) regulating AI companion models — mandatory crisis-referral protocols when users express suicidal ideation, self-harm, or physical harm …

PendingJanuary 1, 2027

Washington

2 laws

Companion chatbot safety

ESHB 2225 (AI companion chatbots)Washington chapter governing AI companion chatbots.

PendingJanuary 1, 2027

AI content transparency

E2SHB 1170 (AI disclosures and content provenance)Washington act creating two chapters that commence together on February 1, 2027.

PendingFebruary 1, 2027

10 of these 17 are not yet in force and are labelled as such rather than folded in with the rest. Commencement labels are quoted from the instrument, including the awkward ones — California’s transparency statute commences in three stages across three years, and flattening that into a single date would be tidier and wrong.

What the map cannot tell you

Three questions it raises and does not answer.

Each is a determination, and each is the reason this page is free.

Whether the statute reaches you at all

Most carry entity thresholds, and several reach developers rather than deployers. Being in the state is the first gate, not the only one.

Which of your tools is a companion chatbot

The definitions are narrower than the headlines and they differ between states. A support assistant and a companion product can be the same software with different settings, and the statutes notice.

What you actually have to do about it

Disclosure, crisis protocols, safeguards for minors, incident reporting and audit duties are separate obligations with separate deadlines. The map shows that a duty exists; the report shows which ones, and when.

Start free See every instrument

State by state

The patchwork is the product problem.

12 states today, and the number has only moved one direction. A determination tells you which of them reaches this deployment — and records the ones that were checked and did not.

Start free Talk to us