SB 3444, known as the Artificial Intelligence Safety Act, is a Illinois state bill that would shield AI labs from liability in cases where AI models are used to cause critical harm. It defines "critical harms" as events such as the death or serious injury of 100 or more people, at least $1 billion in property damage, or a bad actor using AI to develop a chemical, biological, radiological, or nuclear weapon. Coverage under the bill is tied to a model's training expense: any system built on more than $100 million in compute qualifies as a frontier model, a bar that would rope in the country's biggest AI developers, among them OpenAI, Google, Anthropic, xAI, and Meta.