Illinois Moves Forward With AI Safety Laws for Large AI Companies

Illinois lawmakers have introduced a set of bills focused on artificial intelligence safety and holding developers accountable. The centerpiece is Senate Bill 3312, which targets the largest companies building cutting-edge AI systems. This is one of the most detailed state-level attempts at AI regulation in the United States.

What's in the Main Bill

Senate Bill 3312, led by Senator Mary Edly-Allen, creates the Artificial Intelligence Safety Measures Act. It defines what counts as a "frontier model" — roughly, the most advanced AI systems that carry the highest risks — and spells out who qualifies as a large frontier developer.

The bill requires these large companies to adopt and publish a plan for managing serious risks, being transparent about how they operate, and protecting against cyberattacks. The approach is narrower than California's failed attempt at similar rules (SB 1001), which faced pushback from the tech industry over cost and feasibility. Illinois is focusing mainly on the biggest developers rather than trying to regulate the entire AI development ecosystem.

Who Else Gets Covered

Several companion bills round out the package. Senate Bill 3384 creates rules for what are called "companion models" — though the exact requirements will be hammered out in committee.

But Senate Bill 3444 does something quite different: it shields frontier AI developers from lawsuits if their systems cause serious harm. This is a major shift from how product liability usually works. It reflects the fundamental challenge of assigning blame when an AI system behaves in unexpected ways. So while one bill requires safety measures, another limits what happens if those measures fail — a tension that suggests lawmakers are still figuring out how traditional legal rules apply to AI.

Protecting Employees Who Speak Up

Senate Bill 315 updates Illinois's whistleblower protections to cover employees who report violations of the new AI safety rules. This acknowledges that safety problems inside an AI lab might not be visible to outside regulators. Internal workers may be the first to spot when safety requirements aren't being met.

Building on Past AI Rules

Illinois has already started regulating AI in specific areas. In 2024, the state passed HB 3773, which amended employment law to address bias in hiring algorithms. That gives the state some experience with AI regulation.

Representative Eric Sorensen also introduced a separate bill in January 2024 targeting deepfakes and manipulated videos — another example of Illinois taking a multipronged approach to AI-related risks.

What Still Needs to Be Worked Out

Several key details remain unclear and will likely be sorted out as the bills move through committee. The definition of "large frontier developer" — essentially, how big does your compute resources need to be to trigger these rules — is still up for debate. Same with what exactly developers need to include in their published safety plans, and how tough the cybersecurity requirements will be.

The bigger pattern here is worth noticing. We have seen this before with earlier waves of transformative technology. In the 1990s, states grappled with how to regulate the internet, testing different approaches until federal rules eventually took shape. The current state-level AI regulation wave looks similar — each state experimenting with different ideas, waiting to see what actually works.

What This Means for Companies and Regulators

By requiring large AI developers to make their safety plans public, Illinois is creating a new type of disclosure. Other states could study these plans to build their own rules. It also gives competitors a window into how other companies think about safety.

The problem is that Illinois hasn't yet spelled out how it will enforce these rules or what penalties apply for breaking them. That will depend heavily on whether the state can hire or train people with enough AI expertise to check whether companies are actually complying. The impact will also depend on whether major AI companies actually do enough business in Illinois to be subject to the rules.

The liability shield in SB3444 could end up being the most important part of this whole package. If it holds up in court, it might make Illinois an attractive place for AI companies to base operations, despite the safety compliance costs. Whether that happens will likely shape where AI companies choose to set up shop in coming years.