U.S. State AI Safety & Regulation Index

Scoring Rubric

Each of the 15 indicators is scored 0–4 based on enacted law, not proposed bills.

No protection

No relevant enacted law, official rule, or formal policy exists for this question.

Weak or indirect protection

The state only addresses the issue vaguely, indirectly, through a proposed bill, or through a very narrow policy. There is little or no enforceable duty.

Partial protection

A real enacted law or official policy exists, but it is limited — applying only to government agencies, one sector, one type of AI use, or general privacy law rather than AI chatbots specifically.

Strong but limited protection

A clear, enacted, and enforceable requirement directly addresses the issue. However, it is limited in at least one major way: narrow coverage, limited youth focus, limited reporting duties, weak implementation requirements, or no ongoing oversight.

Comprehensive protection

A clear, enacted, enforceable, and broad requirement directly addresses the issue and includes strong implementation measures — such as oversight, penalties, audits, reporting, risk assessments, incident response, documentation, or specific safeguards for minors or high-risk AI systems.

15 Indicators across 5 categories

Category 1 — Transparency & Accountability

Q1 Disclosure when interacting with AI chatbots or AI-generated content
Q2 Stronger disclosure in sensitive areas (mental health, healthcare, elections, finance)
Q3 Records, audits, risk reports shared with regulators or the public

Category 2 — Youth Protections

Q4 Special protections for minors using AI chatbots or AI systems
Q5 Limits on sexual, romantic, manipulative, or emotionally dependent chatbot interactions with minors
Q6 Age-appropriate design, parental controls, or special safeguards for minors

Category 3 — Safety Design & Crisis Response

Q7 Companies must assess and reduce foreseeable chatbot or AI harms before or during deployment
Q8 Safe responses required for self-harm, suicide, abuse, or crisis situations
Q9 Testing, red-teaming, risk mitigation, or incident reporting for higher-risk AI

Category 4 — Data Privacy & User Control

Q10 Limits on collection, storage, or sharing of sensitive chatbot or AI data (especially for minors)
Q11 Limits on using chatbot conversations for ads, profiling, or AI training without consent
Q12 Users or parents can access, delete, correct, or control chatbot data

Category 5 — Enforcement & Legal Strength

Q13 Clear agency, attorney general, or regulator has power to enforce the law
Q14 Meaningful penalties, lawsuits, audits, investigations, or complaint processes
Q15 Law is narrow, clear, and constitutionally safer (safety/privacy/design focus, not broad speech control)

U.S. State AI Safety & Regulation Index

Score / 60

State Rankings