OpenAI backs liability bill
Wired reports OpenAI supported an Illinois bill that would limit when AI labs can be held liable, including in some high‑harm scenarios. The move highlights how major labs are actively shaping legal boundaries for AI responsibility as governance debates continue to evolve ((wired.com)).
OpenAI didn’t just comment from the sidelines on an Illinois artificial intelligence bill. WIRED reported that the company testified in favor of Senate Bill 3444, a state proposal that would limit when frontier model developers can be sued after “critical harms.” (wired.com) The bill is not about ordinary chatbot mistakes like a bad summary or a wrong answer. The Illinois General Assembly says Senate Bill 3444 would stop liability for “critical harms” if the developer did not act intentionally or recklessly and if it published safety and transparency documents. (ilga.gov) That means the legal fight is over where to draw the line between a tool maker and the person using the tool. Illinois is already considering separate bills that treat deployers as liable when they make material changes to an artificial intelligence product or intentionally misuse it. (ilga.gov) This is the part that startled people. WIRED says the Illinois bill’s definition of “critical harm” reaches extreme cases, including mass casualties, major infrastructure damage, and financial losses above $500 million. (wired.com) OpenAI’s position fits a broader pattern in artificial intelligence politics. Big model companies have spent the past year arguing that they should have to publish safety plans and security protocols, but should not automatically be on the hook every time someone uses a model to cause damage. (wired.com) Illinois is not starting from zero here. A state task force report delivered in December 2024 said Illinois should balance innovation with “strong safeguards” and keep updating rules as artificial intelligence systems evolve. (ilga.gov) The politics get messier because Illinois lawmakers are moving on several tracks at once. One Senate bill focuses on safety measures for frontier developers, while another focuses on product liability rules for developers and deployers. (ilga.gov 1) (ilga.gov 2) So the real question is no longer whether artificial intelligence will be regulated. The question is whether the law will treat model makers like car manufacturers, software platforms, or something in between, and OpenAI is now openly trying to shape that answer in Springfield. (wired.com)
