Gov. JB Pritzker signed artificial intelligence legislation modeled after similar bills in California and New York on Monday, furthering calls for a state-led national framework rather than federal regulations.
“Congress and the president should pass similar legislation, but they have been unwilling to do so because many are dependent on special interests that benefit from the lack of regulation in the industry,” Pritzker said before signing the bill. “We can work together to create thoughtful guardrails that benefit both industry and the public, or we can allow a handful of actors to evade responsibility and shift the costs and harms onto ordinary people. Illinois has chosen our path.”
Senate Bill 315also known as the Artificial Intelligence Safety Measures Act, increases transparency and accountability requirements for the largest artificial intelligence models – those that generate more than $500 million in annual revenue and are trained using massive computing power.
The bill mirrors California’s SB-53 and New York Responsible AI Safety and Education Acteach of which was signed at the end of 2025. It sets new reporting standards for the possibility that the AI model could be used for large-scale harm, such as helping users create a chemical, biological or nuclear weapon or commit cyberattacks.
Senate sponsor Sen. Mary Edly-Allen, D-Libertyville, said it is urgent for states to protect themselves from these potential harms.
“We are not prepared to wait for Congress to act,” Edly-Allen said. “There’s an old saying: Give a man a fish, he’ll eat for a day. Teach him to fish, he’ll eat for a lifetime. However, teach the AI to fish and it could empty the whole river trying to figure out how to do it.”
Although the three states represent only about 20% of the national population, lawmakers estimate they represent about 40% of the U.S. AI market, effectively creating a national standard.
Table of Contents
New guard rails
The new law requires model developers to publish an AI framework that outlines how the developer identifies and assesses a “disaster risk,” defined as the likelihood of incidents that could result in the death or serious injury of more than 50 people or more than $1 million in property damage.
Developers must also report any incidents that could cause harm to the state within 72 hours of the incident being discovered, or within 24 hours if there is an imminent risk of death or serious bodily harm.
The bill’s sponsor, Rep. Daniel Didech, D-Buffalo Grove, said the harms the bill will address are not theoretical.
Sign up for Morning Wire:
Our flagship newsletter breaks down the day’s biggest headlines.
“We’ve already seen the first AI-inspired mass shootings. We’ve already seen AI systems used to attack a municipal water and drainage utility,” Didech said.
He also referred to the example of Anthropics Mythos model, as the company said too powerful a cyber weapon to make it accessible to the public. Anthropic supported the Illinois bill and had representatives present at the signing on Monday.
“Every transformative technology in our history, from automobiles to electricity to air travel, has brought enormous benefits while introducing real risks, and in each case the government has responded, not by banning the technology or taking a reckless stance, but by putting safeguards in place so that ordinary people can have confidence that these technologies are safe,” Didech said.
Illinois’ version, similar in many ways to the standards set in New York and California, adds a first-in-the-nation requirement for mandatory annual third-party audits; New York’s version required only a single independent review when developers became large enough to qualify under the law.
During debate in the General Assembly, the third-party audit provision was a point of contention for some industry stakeholders, including TechNet, a coalition of technology executives from across the industry.
“We remain concerned that Illinois would effectively require private actors to make highly subjective decisions requiring AI security compliance without established national standards, certifications, or clear regulatory guardrails,” TechNet representative Ninia Linero told the committee on May 20.
OpenAI and Anthropic both supported the bill as it made its way through the Illinois General Assembly, and it passed with broad bipartisan support in both chambers; only five Republican senators voted against it, namely approved unanimously in the house.
Although major developers have pushed for a federal framework rather than an inconsistent patchwork of government regulations, Caitlin Niedermeyer of OpenAI’s Global Affairs told the Senate AI and Social Media Committee in April that OpenAI was open to a coordinated government-led approach.
“While we have been very clear that the federal government remains well-positioned to lead on border security because it has the resources, the expertise and the institutions, we also see the strong position that Illinois, but also California and New York, can actually take a leadership position in advancing aligned frameworks that we believe can absolutely help create a de facto national direction,” Niedermeyer told the committee.
What’s on the AI horizon?
Companies that violate this will be subject to civil penalties from the Attorney General’s Office of up to $1 million for the first violation and up to $3 million for subsequent violations.
But lawmakers and advocates assume they will continue to address the issue of AI in the future. For example, Didech identified medical care and education as likely frontiers that require further assessment of AI’s risks to public safety.
Scott Wisor, policy director for Secure AI, was one of the advocates who helped shape the Illinois bill. A stronger external assessment of the risks posed by the models and an assessment of when they are ready for publication would be the next step towards greater transparency and accountability, he said.
“Right now the assessment in this bill is: Are you sticking to your security framework? Because assuming you have a security framework, just like, ‘We’re doing A, B, C and D,’ then you do that, the assessor confirms it, and yet it’s still a risky thing to have out in the world,” Wisor said.
“So this is a big step forward, but I think we can do more,” Wisor said.
The Illinois law goes into effect on January 1, 2028.
___
This story was originally published by Capitol News Illinois and distributed through a partnership with The Associated Press.
https://apnews.com/article/landmark-ai-regulations-illinois-statedriven-national-standard-8cfd932ef028fb3784b844f325bf5a10
