The Illinois House Representatives on Wednesday passed a bill requiring cutting-edge AI labs like OpenAI, Anthropic and Google DeepMind to have their safety practices audited by a third party. If the law passes, AI security experts tell WIRED, it would be the first in the country. check the power of big AI companies.
The bill, SB 315, now heads to Governor JB Pritzker’s desk. In a job On social media Wednesday, Pritzker said he plans to sign the bill, citing the need to hold big tech accountable.
Since Congress has yet to pass meaningful legislation on AI safety, state lawmakers have happily rallied in recent years to promote bills that show their constituents they’re keeping Silicon Valley in check. As AI tools become more and more popular and the companies behind them are rushing towards massive IPOspolls show that American voters are seeking more regulation on AI.
As a result, security advocates and tech companies have focused on state legislatures as the main battleground over what these laws should look like. OpenAI’s Head of Global Affairs, Chris Lehane, said to WIRED Last week, the company’s AI policy is now geared toward passing a series of similar state laws.
California and New York have the strictest AI security laws, requiring tech companies to provide information on guardrail models and publish reports on security incidents as they occur. The Illinois bill goes even further, requiring independent auditors to verify that an AI lab is meeting its own safety standards. Previously, no independent body was required to hold an AI lab accountable for its own security claims.
“We’re in a situation where AI companies are grading their own homework,” says Scott Wisor, policy director at Secure AI Project, a nonprofit that supports SB 315. “If SB 315 becomes law, Illinois would require an independent auditor to verify whether AI labs are actually meeting their security commitments.”
Wisor says it is generally anticipated that under SB 315, AI labs could use the Big Four accounting and auditing firms (Deloitte, EY, KPMG and PwC) to audit their security practices. He also says it’s possible that AI labs could exploit members of the AI Evaluators Forum— a coalition of small research organizations, including METR, Transluce and Averi — to assess compliance with safety standards.
Illinois state Rep. Daniel Didech, sponsor of SB 315, told WIRED that state legislatures play an important role in shaping U.S. AI policy and serving as a testing ground for any federal laws that might be passed in the future. “Laws like this create a world where it’s more likely the federal government will pass something,” Didech says.
Business interests
Illinois has become a major arena in the ongoing fight over the nation’s AI laws. OpenAI previously supported a bill in Illinois, it would allow AI labs to avoid liability if their models caused catastrophic harm. However, Lehane has since stated that the company widespread support for the bill was an oversightand he never supported liability protection in the bill. Most recently, OpenAI approved SB 315.
“The Illinois General Assembly has shown true bipartisan leadership in advancing SB 315 and developing a thoughtful framework for AI border security. As AI systems become more capable, clear expectations for security, transparency, incident reporting and accountability are important,” Lehane said in a statement to WIRED.
Anthropic says it is the first AI lab to support SB 315 and thanks the lawmakers who introduced it. Cesar Fernandez, head of U.S. state and local government relations at Anthropic, told WIRED in a statement that SB 315 will help “set a benchmark that all major AI developers are expected to meet.”
However, other Silicon Valley business groups opposed the legislation. The Chamber of Progress, a trade group that lists Google, Apple, Amazon and Andreessen Horowitz as partners, sent a letter to Illinois lawmakers Wednesday morning, asking them to oppose SB 315. The group’s founder and CEO, Adam Kovacevich, told WIRED that the Illinois AI bill “would force companies to expose their sensitive systems to untested auditors in a regulatory regime that has only accountability and no standards.”
During his second term, President Donald Trump signed several executive orders aimed at easing government-imposed AI regulations. Biden administration And state legislaturesciting the need to avoid “fragmented” regulations that could hinder the United States in its race against China. Last week, President Trump canceled the planned signing of a decreesaying he didn’t want to do anything that could blunt America’s competitive advantage.