Illinois Becomes First State to Require Safety Audits for Big AI Companies

Illinois is now the first state in the country to require large artificial intelligence companies to go through independent safety audits. Gov. JB Pritzker signed the new law on Monday in Chicago. The law is meant to make sure that powerful AI technology is being developed safely and that companies are held responsible for the risks their products may create. Illinois lawmakers said they felt they had to act because the federal government has not passed its own rules for AI.

The new law targets the biggest AI companies — those that make more than $500 million in revenue each year. These companies will have to publicly explain how their AI products could cause serious harm and what steps they are taking to prevent it. They will also have to report major safety problems to the state within 72 hours of learning about them. These rules will go into effect on January 1, 2028.

Every year, these large AI companies will be required to hire an outside, independent expert to audit — or carefully review — whether they are following the safety rules. The outside experts must have real knowledge of advanced AI systems, sometimes called 'frontier AI models.' This kind of outside check is what makes the Illinois law unique compared to laws passed in other states. No other state has required this kind of independent review before.

The law also protects workers who speak up about safety problems. Companies cannot punish employees who report concerns to state or federal authorities. Workers must also be given a way to make anonymous reports inside their own companies if they believe the company's actions are putting people in danger. Fines for breaking the law can reach $1 million for a first offense and $3 million for repeat violations.

Gov. Pritzker said that states need to lead on AI rules when the federal government is not acting. He said that careful rules are needed to make sure new technology grows in a way that is safe and lasting. Illinois House Speaker Emanuel 'Chris' Welch said decisions about AI are too important to be left only to tech companies that move quickly without thinking about consequences.

State Sen. Mary Edly-Allen, the bill's lead sponsor in the Senate, said lawmakers need to learn from past mistakes. 'If we got social media wrong, and we did, we cannot afford to get AI wrong at an even greater scale,' she said. She and other lawmakers believe that AI is a much bigger and faster-moving technology than social media, which means the risks are even greater.

Some of the biggest AI companies actually supported the new law. Anthropic, the company that makes the Claude chatbot, backed the bill and said the law makes safety steps a requirement for everyone. OpenAI, the company behind ChatGPT, also supported the law. This broad support helped the bill pass with votes from both Democrats and Republicans.

Not every AI bill made it through the legislature before lawmakers took their summer break. One bill would have required AI companies to detect when a user might be at risk of self-harm and connect them to a crisis service. Another bill that fell short would have stopped companies from selling people's most sensitive personal data without first giving them a chance to say no. Supporters of those bills say they hope to try again when lawmakers return.