Illinois Governor Signs AI Safety Law Requiring Yearly LLM Audits
Illinois just told the biggest names in artificial intelligence (AI) that the rules are changing
On Monday (July 6), Gov. JB Pritzker signed a law that forces the largest AI companies to prove their most powerful systems are safe before those systems reach the public. The state isn’t waiting for Washington to act. It’s building its own set of guardrails, and it’s betting that other states will follow
According to the Associated Press (AP), the new law is called the Artificial Intelligence Safety Measures Act, also known as Senate Bill 315. The statute raises transparency and accountability standards for the largest AI models, specifically those that pull in more than $500 million in annual revenue and run on enormous computing power. The AP reported that Illinois modeled its bill after laws already signed in California and New York in late 2025
Together, these three states hold only about 20% of the country’s population, but lawmakers estimate they cover close to 40% of the U.S. AI market. That gives them the muscle to set a national standard without any help from Congress
The law targets what it calls catastrophic risk. Think of it like fire codes for a skyscraper. Nobody expects the building to burn, but the rules exist because the damage would be severe if it did. Under the new Illinois law, that means any incident that could kill or seriously injure more than 50 people, or cause more than $1 million in property damage. Companies must publish a framework explaining how they spot and manage these dangers. They also have to report harmful incidents to the state within 72 hours, or within 24 hours if the threat is immediate.
Supporters argued the risks are already real, not hypothetical. Rep. Daniel Didech, the bill’s House sponsor, made that case directly
The AP quoted Didech as saying: ”We have already seen the first AI-inspired mass shooting. We have already seen AI systems utilized to attack a municipal water and drainage utility.” He compared AI to earlier breakthroughs like cars, electricity and air travel. Each one delivered huge benefits while carrying serious dangers, and in each case the government responded by building safety measures rather than banning the technology or ignoring it
Illinois added one requirement that goes farther than its neighbors. The state now mandates yearly third-party audits, the first such requirement in the nation, according to the AP. New York, by contrast, only requires a single independent audit once a developer grew large enough to fall under the law. That Illinois audit provision sparked pushback during debate. TechNet, a coalition of technology executives, warned that the state would be asking private companies to make highly subjective safety calls without clear national standards to guide them.
Even so, the bill drew wide bipartisan support. It passed the House unanimously and lost only five Republican votes in the State Senate. Both OpenAI and Anthropic backed the measure as it moved through the legislature, even though large developers had pushed for a single federal framework instead of a patchwork of state rules
So, what comes next? Companies that break the law face civil penalties of up to $1 million for a first offense and up to $3 million for later ones, enforced by the attorney general’s office. Lawmakers say they aren’t finished. According to the AP, Didech pointed to medical care and education as areas that will likely need their own rules down the road. Advocates who helped shape the bill want future versions to include outside experts who can judge whether a model is truly ready for release, not just whether a company followed its own safety plan.
The Illinois law takes effect on Jan. 1, 2028. That gives companies time to prepare, and it gives other states time to decide whether to copy the blueprint
