Most companies lobby to limit liability for minor harms. OpenAI is lobbying to limit liability for catastrophic ones.
The company testified in favor of an Illinois bill that would restrict when AI labs can be held legally responsible for damages - including in cases the bill categorizes as "critical harm." That category covers scenarios like mass casualties and large-scale financial disasters, according to Wired's reporting.
The bill's framing inverts the usual logic of liability law, which typically holds that the more serious the potential harm, the greater the obligation to prevent it. Under this proposal, companies that build the most powerful AI systems could face reduced legal exposure even when those systems contribute to the worst outcomes.
OpenAI's position isn't surprising in isolation - tech companies routinely lobby for favorable regulations. But the specific scope of the carve-out is worth parsing. Protecting ChatGPT and similar products from lawsuits over defamatory outputs is one thing. Shielding model providers from suits connected to mass casualties is a different category of ask entirely.
Illinois is one of several states moving ahead with AI-specific legislation as federal action remains gridlocked. The outcomes of bills like this one will set precedents for how AI companies are held accountable before any federal framework exists. If a liability shield of this scope passes at the state level, expect it to serve as a template for similar pushes in other states.