With the White House and Congress stalled on AI legislation, the battlefield has moved to state capitols — and America’s two leading frontier labs have split wide open there. Per Wired and Business Insider reporting, Anthropic and OpenAI give opposite answers to “what should state AI safety laws look like,” and most state-level battles since summer 2026 have fought along that fault line.
Two Strategies: The Ratchet vs the Uniform Template
Anthropic’s play is ratcheting: back progressively stricter bills state by state, so strict standards ratchet upward against each other. State policy lead Cesar Fernandez: “We’re looking for legislation that meaningfully raises the bar on safety,” and action “can’t wait for action in Washington.” OpenAI, led by top lobbyist Chris Lehane, runs “reverse federalism”: replicate the same (California-style) framework across states into a de facto national standard. Spokesperson Liz Bourgeois argues it “gives the public clearer protections” and spares developers conflicting rules.
Illinois SB 3444: Collision Over a Liability Shield
The sharpest clash is in Illinois. Senator Bill Cunningham’s SB 3444 would shield AI labs from liability when their systems are used to cause large-scale harm — mass casualties or more than $1 billion in property damage — as long as the lab drafted and published its own safety framework on its website. OpenAI backs it, arguing it reduces serious-harm risk while letting the technology reach Illinois businesses.
Anthropic lobbied heavily to amend or kill the bill. Fernandez’s critique is blunt: transparency laws should “ensure public safety and accountability,” not provide “a get-out-of-jail-free card against all liability.” Governor JB Pritzker opposes it too — he “does not believe big tech companies should ever be given a full shield.” Secure AI Project’s Thomas Woodside said the bill “would take the extreme step of nearly eliminating liability for severe harms.” Policy experts give SB 3444 a remote chance of passing, but the fight exposed the political gulf between the two labs.
State-by-State Scoreboard
- California: its 2025 law — the first US state law regulating advanced models — was endorsed only by Anthropic among major labs; Anthropic treats it as a floor, not a ceiling
- New York: per Business Insider, it was OpenAI’s lobbyists (not Anthropic’s) who pressed Governor Hochul late last year to amend the state’s bill toward the California framework; Anthropic had backed the stronger original
- Illinois: Pritzker signed the audit-direction bill (SB 3261) in July 2026, requiring annual independent third-party audits of safety plans — a first-of-its-kind mandate that both OpenAI and Anthropic backed
- Massachusetts: in late June, Anthropic endorsed provisions in an economic development bond bill it calls the nation’s toughest — independent evaluators assessing catastrophic risks (such as bioweapons assistance) and state attorney general enforcement; OpenAI is still reviewing
Money and Motives
The state war is getting expensive: dueling super PAC networks tied to each company have spent tens of millions, and Anthropic began donating directly to California legislators in June. Critics — some tied to the Trump administration and venture capital — frame Anthropic’s push as regulatory capture designed to hobble rivals.
Business Insider also notes a technical motive: Anthropic’s internal testing found Claude Mythos could exploit flaws in every major operating system, and its public sibling Fable triggered Trump administration export controls. In effect, Anthropic is using its own strongest model’s test results as first-hand evidence for why strict rules are needed.
What to Watch
For teams tracking AI governance, two things matter: whether Illinois’s third-party audit model gets copied by other states (the ratchet strategy’s most concrete output so far), and whether OpenAI’s uniform template wins its first big state. Whichever model wins will become the draft text of future federal legislation.
Sources
- Anthropic Opposes the Extreme AI Liability Bill That OpenAI Backed — Wired
- Inside Anthropic’s State-by-State Plan to Ratchet up AI Rules — Business Insider
AI-assisted summary compiled from the sources above, reviewed by a human before publishing.