Illinois just signed a frontier AI law—here’s what it changes
Illinois Governor JB Pritzker has signed a new artificial intelligence safety law that targets the biggest “frontier” AI developers, requiring transparency about risk controls and regular independent audits. Supporters say it’s meant to set accountability guardrails as powerful models spread; critics warn it could clash with emerging national standards.
Key Takeaways
- Pritzker signed the Artificial Intelligence Safety Measures Act to regulate leading AI companies’ most advanced models.
- Illinois is positioned as a first-mover on audits, requiring annual third-party reviews of safety plans for the largest labs.
- Companies must evaluate and mitigate catastrophic risks, plus publish a transparency framework and face enforcement by the Illinois attorney general.
- Penalties can reach millions, including up to $1 million for a first violation and up to $3 million for additional violations.
What happened in Illinois, and why does it matter?
In Chicago, Gov. JB Pritzker signed the Artificial Intelligence Safety Measures Act into law, calling for more transparency and accountability as AI use grows, according to ABC7 Chicago. The law is aimed at regulating the most capable models built by the largest companies, rather than everyday software tools.
Why it matters: state governments are increasingly trying to define practical rules for fast-moving AI systems when federal standards are still taking shape. In Illinois’ case, the centerpiece is external verification—moving from “trust us” to “show us” when companies claim their safety plans work.
What does the new law require big AI labs to do?
The measure requires large AI companies to develop risk-mitigation frameworks and undergo yearly independent audits, ABC7 reported. Developers would have to create and publish a transparency framework explaining how the company applies industry standards, measures model capabilities and the chance of catastrophic risk, and identifies and responds to safety incidents.
Bloomberg Law reports the act requires companies developing frontier AI models to evaluate and mitigate catastrophic risks, submit to independent third-party audits, report critical safety incidents within 72 hours, and ensure public disclosures are accurate. Bloomberg Law also describes “catastrophic events” in the law as those resulting in the deaths of at least 50 people or damage of $1 billion or more per incident.
The Hill adds that Illinois is the first state to mandate that the largest AI labs obtain third-party audits of their safety plans, and that the bill mirrors efforts in California and New York while going further with required annual audits for frontier labs above a revenue threshold.
Who is covered, and what are the penalties?
The law is targeted at the biggest developers through its thresholds, including a $500 million revenue benchmark, ABC7 reported. That means the rules are designed for the largest “frontier” model builders—not startups tinkering in a garage.
Enforcement is built in. ABC7 reported the Illinois attorney general would have authority to fine companies that violate the law—up to $1 million for a first violation and up to $3 million for additional violations. Bloomberg Law similarly notes penalties can reach as high as $3 million for repeat violations.
Is everyone on board with Illinois regulating frontier AI?
Not entirely. ABC7 reported that while the bill passed unanimously in the Illinois House and was supported during the legislative process by OpenAI and Anthropic, the third-party audit requirement remains a point of contention for some industry stakeholders, including TechNet (a coalition of tech executives).
Still, supporters argue Illinois is trying to push a credible, enforceable standard: publish what you do, prove it via independent review, and disclose serious incidents quickly. For readers tracking how AI rules are evolving, this is the kind of state-level template other legislatures may copy—or fight.
More on this topic: explore our ongoing coverage in Future Tech & AI Wonders.