AINEWS
Three states just wrote America's de facto AI safety rule
Illinois, California, and New York hold about 40% of the U.S. AI market, and their aligned safety audit laws now function as a federal floor.
Illinois, California, and New York hold about 40% of the U.S. AI market, and their aligned safety audit laws now function as a federal floor.
When Gov. JB Pritzker signed Illinois's Artificial Intelligence Safety Measures Act in Chicago on Monday, he added the third pillar of a national AI safety rule. Congress never wrote it. Federal lawmakers never passed it. The standard exists anyway.
Illinois, California, and New York together hold about 20% of the U.S. population but a share of the U.S. AI market that lawmakers backing the bill put at roughly 40%, per Capitol News Illinois reporting. A frontier model trained once has to clear all three jurisdictions to reach all three markets. Developers don't exit the country's largest customer cluster to dodge safety audits. Three state laws therefore function as a federal floor, built without a federal vote.
SB-315 applies to AI companies operating frontier models that generate more than $500 million in annual revenue, per Capitol News Illinois reading of the bill, and trains on what the enrolled text defines as massive computing power. Covered operators must submit to third-party safety audits of each new frontier release, file pre-deployment notices with the state, and report on catastrophic-harms pathways tied to chemical, biological, radiological, and nuclear weapon assistance and to cyberattacks on critical infrastructure, per the Washington Examiner. The Chicago Tribune called the framework first-in-nation when the bill cleared the General Assembly weeks before Monday's signing.
The framework lines up with California's SB-53 and New York's Responsible AI Safety and Education Act, both signed in late 2025. A single compliance posture now travels across the three biggest state AI markets: one set of auditors, one filing schedule, one catastrophic-risk disclosure form. WIRED reported Illinois's framework as America's strongest state-level AI safety regime.
At the Chicago signing ceremony, Pritzker described federal inaction as 'driven by special interests that profit from the industry having no regulation' and said Illinois 'has chosen our path.' Its Senate sponsor, state Sen. Mary Edly-Allen (D-Libertyville), put the urgency more sharply in NPR Illinois reporting: 'We are not willing to wait for Congress to act.' The line targets Washington as much as the rest of Springfield, the same explicit posture that turned California's SB-53 into a national talking point and pushed New York's measure across the finish line last fall.
For developers, the practical effect is one audit program crossing three regulators. A new frontier release priced for Illinois, California, and New York becomes the release sold to the rest of the country by default, because no operator trains and audits a second model for Texas, Florida, or Ohio. For consumers in those other 47 states, the same disclosures, pre-deployment checks, and catastrophic-risk reports apply to the frontier models they end up using anyway. For federal lawmakers, the choice narrows to preempting the three-state rule with a weaker national standard, sitting out the regulation entirely, or accepting the floor the states have already written.
The state floor is now set. Watch where the friction lands first: a developer lawsuit in federal court, a Congressional move to preempt state AI rules, or a divergence between California's, Illinois's, and New York's audit interpretations that complicates the single-posture strategy the three laws were designed to enable.