Illinois passes the strictest AI safety law in the US, with mandatory audits

Illinois Governor JB Pritzker signed the Artificial Intelligence Safety Measures Act on Monday, a law that puts the state at the forefront of state-level AI regulation in the United States. According to a statement from the governor's office, the law requires "developers of the largest advanced AI systems"—a category that in practice targets companies like OpenAI, Anthropic or Google—to publish information about their safety practices, report any significant security incidents, and comply with a set of regulatory compliance processes. The law also creates a channel for AI lab employees to safely and confidentially report potential risks without fear of retaliation.

Much of the law's content follows the line of what has already been passed in California and New York, but Illinois claims a "first in the nation" distinction for itself: it is the only state that will require independent, periodic security audits carried out by third parties. This obligation will apply to AI companies reporting more than $500 million in revenue, and will take effect on January 1, 2028.

One of the most notable aspects of the case is that, unlike other regulatory battles over AI, this law has received explicit backing from the very companies it aims to regulate. Both OpenAI and Anthropic endorsed it as it moved through the state legislature. This contrasts with a previous episode in which the two companies turned Illinois into a battleground: OpenAI had pushed a separate bill that would have shielded AI companies from large-scale damage lawsuits caused by their systems, a measure Anthropic opposed and which, according to the article, appears to have stalled completely.

The article notes that there was no significant opposition to the Artificial Intelligence Safety Measures Act, likely because its content closely resembles the laws already in effect in California and New York. Given the lack of a meaningful federal framework for AI safety in the United States, companies like OpenAI have chosen to lobby at the state level with the goal of building a set of regulations similar enough to each other to ease nationwide compliance. The company had already actively participated in the legislative processes in California and New York, and can now add Illinois as another step in that strategy of stitching together, state by state, a kind of uniform regulatory patchwork that is more manageable for it than facing a mosaic of disparate rules.

Taken together, the Illinois case illustrates an increasingly visible dynamic in AI governance in the United States: in the absence of federal regulation, states are setting the pace, and the major AI companies themselves—far from always resisting regulation—sometimes prefer to support predictable laws harmonized with other states rather than risk a patchwork of contradictory rules or tougher frameworks imposed without their input.