Reports · 2025-2026 sessions
How is each state regulating artificial intelligence?
Florida, Montana, and Virginia score highest on USLege's AI-Friendliness Index; Minnesota and Vermont score lowest. Here's the 0-100 score for all 50 states, built from every AI bill USLege has tracked across the 2025-2026 sessions, with the actual sponsors, hearings, and outcomes behind each score.
See also: our deep dive on landmark AI bills that passed or died in eight high-volume states.
By Laura Davis, Co-Founder & Chief Business Officer at USLege.
Methodology: each state is scored 0-100 as an AI-Friendliness Index across three equally-weighted factors, each independently rated 1-5 against every substantive 2025-2026 session AI-policy bill USLege's database returned for that state: innovation and enabling measures, how light the private-sector mandate load is, and enforcement restraint (fewer penalties and less liability exposure). The net score is the sum of the three factor ratings out of 15, scaled to 100: the visible score breakdown on every state's card is that same arithmetic, not a separate judgment. Verified against USLege's live legislative database as of September 30, 2026. All 50 states are scored this cycle: Nevada, shown with no score in the prior snapshot because its legislature was not in regular session, held its 2025 regular session on schedule and has genuine AI-policy bill activity.
Every state is legislating on artificial intelligence right now, but how each one balances innovation against restriction varies enormously. This report scores all 50 states on that balance, then shows the actual bills, sponsors, and committee hearings moving each state's law, not just a policy summary, but who is pushing to enable AI and who is pushing to restrict it, in their own words on the record. Sponsorship in this data set follows a clear pattern: bills that restrict AI use skew toward Democratic sponsors, while bills that enable or incentivize AI adoption skew toward Republican sponsors, though many bills in both directions have bipartisan support.
49
states scored
24–78
score range
241
bills analyzed
55
signed into law
Lower
Higher
No data
Top 12 most AI-friendly states
#1 of 50
Florida FL
Very Friendly · Flagship: SB 482
87
AI-Friendly Index
Why this score: Florida's legislature has repeatedly attempted a comprehensive 'Artificial Intelligence Bill of Rights' (SB 482/HB 1395 in the 2026 Regular Session, re-filed as SB 2 in the 2026 Special Session) restricting companion chatbots, AI-generated likeness use, and foreign-linked AI contracts, plus standalone companion-chatbot bills (HB 659, SB 1344/SB 1346) and an AI-in-education mandate (SB 1194) — but every one of these died in committee or 'in messages' across three consecutive sessions, leaving Florida with no enacted AI-specific mandate or enforcement regime from this cycle despite persistent restrictive intent.
Score breakdown Net index 87/100
Innovation / enabling measures
Light private-sector mandate load
Enforcement restraint (less penalty/liability)
Verified against 22 bill(s) tracked in USLege for the 2025-2026 session, as of September 30, 2026
#2 of 50
Montana MT
Very Friendly · Flagship: SB212
87
AI-Friendly Index
Why this score: Montana's 2025 regular session (the only regular session of this biennium; the legislature does not meet in 2026) enacted a strongly pro-innovation AI framework and let the one heavy-mandate bill die. SB212 (Right to Compute Act, chaptered 4/17/2025) declares that government restrictions on lawful computing must be narrowly tailored and imposes only light, federally-aligned risk-management duties on AI-controlled critical infrastructure. SB67 (chaptered 5/8/2025) authorizes autonomous-vehicle operation statewide. HB178 (chaptered 5/8/2025) and SB25 (chaptered 5/13/2025) restrict government AI misuse and require deepfake disclosure in elections -- narrow, not private-sector compliance mandates. SB413 (chaptered) criminalizes AI-generated nonconsensual explicit media. The one bill imposing private-sector AI mandates, HB556 (AI restrictions on health-insurance utilization review), died in process, as did SB452's AI-disclosure mandate for online media.
Score breakdown Net index 87/100
Innovation / enabling measures
Light private-sector mandate load
Enforcement restraint (less penalty/liability)
Verified against 30 bill(s) tracked in USLege for the 2025-2026 session, as of September 30, 2026
#3 of 50
Virginia VA
Very Friendly · Flagship: SB 365
87
AI-Friendly Index
Why this score: Virginia's heavy AI-regulation proposals have repeatedly failed to become law: the FAIR AI Act (SB 365/HB 713) was continued to next session and left in Appropriations in 2026, the High-Risk AI public-bodies framework (HB 2046/SB 1214) failed/died in 2025, and automated-decision-system discrimination and employment bills (HB 999, HB 1514) were left in Appropriations in 2026. What Virginia has actually enacted is light-touch and education-focused: SB 394 establishes VDOE guidance and an AI-in-Education innovation pilot program, and HJ 32 directs a JLARC study of higher-ed AI policies. With no binding private-sector AI mandate or penalty regime enacted across two sessions, Virginia's real regulatory footprint on AI remains minimal despite recurring restrictive proposals.
Score breakdown Net index 87/100
Innovation / enabling measures
Light private-sector mandate load
Enforcement restraint (less penalty/liability)
Verified against 20 bill(s) tracked in USLege for the 2025-2026 session, as of September 30, 2026
#4 of 50
Delaware DE
Very Friendly · Flagship: HJR 7
80
AI-Friendly Index
Why this score: Delaware enacted HJR 7, directing its AI Commission to build a regulatory sandbox for testing agentic AI (signed July 2025), and HB 191, a narrow measure barring AI systems from being licensed as medical professionals (signed April 2026) — both light-touch. Heavier bills remain pending: HB 306 (chatbot-disclosure mandate with civil remedies) and HB 453 (ban on AI-driven surveillance pricing) have not been enacted, leaving Delaware's current binding AI-specific law limited to the enabling sandbox and a narrow licensing carve-out.
Score breakdown Net index 80/100
Innovation / enabling measures
Light private-sector mandate load
Enforcement restraint (less penalty/liability)
Verified against 7 bill(s) tracked in USLege for the 2025-2026 session, as of September 30, 2026
#5 of 50
Mississippi MS
Very Friendly · Flagship: SB2426
80
AI-Friendly Index
Why this score: Mississippi enacted two light-touch AI measures rather than binding mandates: SB2426 (2025, approved by Governor 3/18/2025) creates the Artificial Intelligence Regulation (AIR) Task Force, explicitly framed as a regulatory sandbox to balance innovation with oversight through 2027; HB1723 (2026, approved by Governor 3/9/2026, effective 7/1/2026) merely codifies a legal definition of 'artificial intelligence' with no new duties. Restrictive proposals -- HB1720/HB1048 (AI therapy bans), SB2354 (AI Fraud and Accountability Act, private right of action), SB2642/SB2050 (political-ad AI disclosure), HB1810 (AI-displacement tax) -- all died in committee or on the calendar. No enacted bill imposes developer/deployer mandates, audits, or private rights of action.
Score breakdown Net index 80/100
Innovation / enabling measures
Light private-sector mandate load
Enforcement restraint (less penalty/liability)
Verified against 30 bill(s) tracked in USLege for the 2025-2026 session, as of September 30, 2026
#6 of 50
North Dakota ND
Very Friendly · Flagship: HB 1448
80
AI-Friendly Index
Why this score: North Dakota enacted a mostly enabling AI/tech agenda: HB 1448 (Advanced Technology Review Committee and Grant Program, filed 5/2/2025) and SB 2256 (NDSU research technology park grant for autonomous mobile equipment) both became law, while a larger AI compute-credits program (HB 1265) failed 13-33. Enacted restrictions are narrow and criminal-law-specific — HB 1167 (political AI-deepfake disclosure) and HB 1429/HB 1613 (robot harassment and police robot/drone surveillance limits) — rather than a broad AI compliance or private-liability regime.
Score breakdown Net index 80/100
Innovation / enabling measures
Light private-sector mandate load
Enforcement restraint (less penalty/liability)
Verified against 26 bill(s) tracked in USLege for the 2025-2026 session, as of September 30, 2026
#7 of 50
Utah UT
Very Friendly · Flagship: HB 320
80
AI-Friendly Index
Why this score: Utah continues to run the nation's most developed AI regulatory-sandbox model: HB 320, Office of Artificial Intelligence Policy Amendments (Governor Signed, 2026), refines regulatory mitigation and joint interpretation agreements for supervised AI testing, while HB 273 and SB 322 extend sandbox concepts into classroom AI use. Enacted consumer-protection layers are targeted rather than sweeping — SB 226 (AI Consumer Protection Amendments, 2025, disclosure duties with a safe-harbor) and HB 452 (mental health chatbot disclosure/privacy rules, 2025) — enforced by the Division of Consumer Protection rather than through private rights of action. A broader frontier-model transparency bill (HB 286, AI Transparency Amendments) and a companion-chatbot safety act (HB 438) both died in 2026 (enacting clause struck), leaving Utah's framework innovation-first with modest, narrowly-scoped mandates.
Score breakdown Net index 80/100
Innovation / enabling measures
Light private-sector mandate load
Enforcement restraint (less penalty/liability)
Verified against 25 bill(s) tracked in USLege for the 2025-2026 session, as of September 30, 2026
#8 of 50
New Mexico NM
Very Friendly
73
AI-Friendly Index
Why this score: New Mexico's entire AI-restriction slate died in committee this cycle: SB 68 (AI Government Use Act), HB 28 (AI Transparency Act), HB 60 (broad high-risk AI Act), HB 141 (AI Accountability/synthetic-content labeling), HB 174 (Chatbot Safety Act) and HB 401 (Synthetic Content Accountability Act) were all Action Postponed Indefinitely. What did move was economic-development-oriented: HB 20 (Technology and Innovation Division, signed Ch. 133, Apr. 9 2025) and SB 177 (advanced-tech fund transfers including AI). With no AI-specific mandate or enforcement regime actually enacted, the state is light-touch largely by default.
Score breakdown Net index 73/100
Innovation / enabling measures
Light private-sector mandate load
Enforcement restraint (less penalty/liability)
Verified against 30 bill(s) tracked in USLege for the 2025-2026 session, as of September 30, 2026
#9 of 50
South Dakota SD
Very Friendly · Flagship: SB 164
73
AI-Friendly Index
Why this score: South Dakota's substantive AI bills mostly stalled in the 2026 session: SB 169 (AI utilization-review mandates for health carriers) and HB 1144 (restricting AI in therapy) were withdrawn, HB 1125 (AI study taskforce) failed on the House floor 26-43, and SB 168 (minor chatbot safeguards) and SB 170 (chatbot disclosure) were deferred to the 41st legislative day (effectively killed). The only enacted AI-adjacent law is SB 164, signed by the Governor in 2025, criminalizing political deepfakes within 90 days of an election (Class 1 misdemeanor plus civil liability). With no comprehensive AI framework enacted and no sandbox or funding measures advancing, South Dakota currently sits close to a light-touch middle ground.
Score breakdown Net index 73/100
Innovation / enabling measures
Light private-sector mandate load
Enforcement restraint (less penalty/liability)
Verified against 6 bill(s) tracked in USLege for the 2025-2026 session, as of September 30, 2026
#10 of 50
Texas TX
Very Friendly · Flagship: HB 149
73
AI-Friendly Index
Why this score: Texas enacted HB 149, the Texas Responsible AI Governance Act (TRAIGA), effective 1/1/2026, which prohibits specific harmful AI uses (government social scoring, unlawful discrimination, manipulation) and requires consumer disclosures, but pairs this with an AI regulatory sandbox program, a new Texas Artificial Intelligence Council, and workforce development grants, with enforcement reserved to the Attorney General and no private right of action. Companion enabling measures — HB 3512 (AI/cybersecurity training for state employees, effective 9/1/25), HB 2818/SB 1700 (AI Division within the Department of Information Resources, effective 9/1/25) — reinforce a government-modernization, sandbox-first posture. Narrower proposals (automated employment decision tools, biometric AI disclosure, AI in health care) remain pending, unenacted.
Score breakdown Net index 73/100
Innovation / enabling measures
Light private-sector mandate load
Enforcement restraint (less penalty/liability)
Verified against 20 bill(s) tracked in USLege for the 2025-2026 session, as of September 30, 2026
#11 of 50
West Virginia WV
Very Friendly · Flagship: HB 3187
73
AI-Friendly Index
Why this score: West Virginia enacted HB 3187 in 2025 (Chapter 38), creating the West Virginia Task Force on Artificial Intelligence to recommend state oversight and study economic opportunities, and separately enacted SB 198 (Chapter 80, 2025), a narrowly-targeted criminal ban on AI-generated child sexual abuse material. The 2026 slate — HB 5205 (Balance Act: AI/tech-use model policies plus optional high-school AI sandbox courses), HCR 4 (Genesis Working Group coordinating AI/HPC/energy economic development), HB 4770 (AI mental-health-care limits), HB 4496 (AI-media disclosure with AG civil penalties), and SB 70 (foreign-adversary AI software ban on state systems) — remained in committee as of the latest action. Enacted mandates so far are narrow and enforcement is limited to a single targeted criminal statute, alongside genuine (if modest) innovation-coordination infrastructure.
Score breakdown Net index 73/100
Innovation / enabling measures
Light private-sector mandate load
Enforcement restraint (less penalty/liability)
Verified against 8 bill(s) tracked in USLege for the 2025-2026 session, as of September 30, 2026
#12 of 50
Wyoming WY
Very Friendly · Flagship: HB 102
73
AI-Friendly Index
Why this score: Wyoming's AI-specific docket is thin: HB 91 (banning government AI social-scoring and warrantless AI biometric identification) died when the House declined to consider it for Committee of the Whole, while HB 102, "Protecting kids from deepfakes and exploitative images," was enacted (Chapter 91, effective 7/1/2026) — it criminalizes AI-generated CSAM and self-harm-promoting AI while explicitly granting developers liability immunity absent knowing intent to enable illegal use. Beyond these two, most "AI"-tagged hits in Wyoming's docket (Digital Taxonomy Act, Cryptographic Frontiers Act, Select Committee on Blockchain and Emerging Technologies) are blockchain/crypto measures rather than AI policy proper. With its one enacted AI law shielding rather than burdening developers, and no comprehensive AI mandate regime proposed, Wyoming remains a light-touch state on AI by default.
Score breakdown Net index 73/100
Innovation / enabling measures
Light private-sector mandate load
Enforcement restraint (less penalty/liability)
Verified against 2 bill(s) tracked in USLege for the 2025-2026 session, as of September 30, 2026
The rest of the country
Unlock the full report
See all 50 states
Tell us who you are to unlock every state card, bill, and sponsor in this report, free, no credit card required.
Track every AI bill in every state, live, with the sponsor, the hearing video, and the vote history behind each one.
Book a demo →Frequently asked
Which states are most friendly to AI companies?
Utah, Idaho, North Dakota, Montana, and Wyoming score highest in this index, weighted toward innovation and enabling measures, a lighter private-sector mandate load, and enforcement restraint.
Which states regulate AI most restrictively?
Illinois, Minnesota, Colorado, Vermont, and Massachusetts score lowest, generally carrying comprehensive private-sector mandates, broader enforcement exposure, and fewer innovation-enabling measures.
How is the AI-Friendliness Index calculated?
Each state is scored 0-100 across three weighted factors: innovation and enabling measures, how light or heavy the private-sector mandate load is, and enforcement restraint, meaning fewer penalties and less liability exposure. The score is built from every 2025-2026 session bill USLege's database returned for the topic, not standing law alone.
Where does this data come from?
Directly from USLege's live legislative tracking database, which indexes bill text, sponsors, actions, and committee hearing video across all 50 states in real time. This snapshot reflects every state verified against USLege's live database as of September 30, 2026, across the 2025-2026 regular sessions.
Why is Nevada shown with no score?
Nevada's legislature meets biennially and was not in regular session during this snapshot, so no current-session AI bill activity is indexed there. It is shown as neutral, not scored, rather than assigned an estimated value.