Illinois mandates independent safety audits for frontier AI models

Illinois Governor J.B. Pritzker confirmed this week he will sign Senate Bill 315 into law, making Illinois the first US state to require independent, third-party audits of frontier artificial intelligence models. The bill passed the Illinois House 110-0 on May 27 and the Senate 52-5 the week prior.

What the law actually requires

SB 315 targets the most capable AI models from the largest companies — specifically those with at least $500 million in revenue and a qualifying computing threshold. Covered developers must create and publicly publish a transparency framework explaining how they assess model capabilities, measure catastrophic risk, and respond to safety incidents.

The law mandates annual independent third-party audits of those safety protocols — a requirement that currently has no parallel at the federal level in the United States. Developers must also report critical safety incidents to Illinois authorities within 72 hours, or within 24 hours if there is an imminent risk of death or serious physical harm. Employees at covered companies gain protected whistleblower status for reporting emerging safety risks internally.

Civil penalties for violations reach up to $3 million per incident, enforced exclusively by the Illinois attorney general. The bill includes no private right of action. Provisions take effect January 1, 2027.

OpenAI and Anthropic supported it

Notably, the two US AI companies whose models would most clearly fall under the law — OpenAI and Anthropic — both supported the legislation. Anthropic's head of state and local government relations, Cesar Fernandez, called SB 315 "a baseline that every leading AI developer is expected to meet." OpenAI's chief of global affairs, Chris Lehane, told Wired the company is actively pushing for similar measures in other states, a strategy analysts interpret as a preference for uniform standards over a patchwork of inconsistent state laws.

Not every industry voice agreed. Adam Kovacevich, CEO of Chamber of Progress — whose members include Google and Apple — argued the law "would force companies to expose sensitive systems to untested auditors in a regulatory regime that is all liability and no standards."

Why states are stepping in

The Illinois bill follows regulatory moves in New York and California in 2025, and arrives days after the Trump administration shelved a federal executive order that would have required voluntary pre-release review of advanced AI models. With Congress having passed no federal AI legislation, state governments are filling the vacuum — a pattern that Illinois House sponsor Rep. Daniel Didech acknowledged is suboptimal. "The states shouldn't be doing this," Didech said. "But Congress has not taken up this issue, and the technology is developing at such a rapid pace that states have had no choice but to step in."

Scott Wisor of the Secure AI Project, which supported the bill, framed the core problem bluntly: without enforcement requirements, "we're in a situation where the AI companies grade their own homework." As reported by Capitol News Illinois, the Illinois framework is being explicitly designed as a potential template for federal action — a testing ground for what enforceable AI oversight might look like at scale.