OpenAI Backs Bill That Would Limit Liability for AI-Enabled Mass Deaths or Financial Disasters

OpenAI, the company behind ChatGPT, has formally testified in support of proposed legislation in the state of Illinois that seeks to establish legal liability shields for artificial intelligence developers. The company's representatives advocated for the bill before Illinois lawmakers in early 2025, arguing that such protections are necessary to foster innovation in the high-stakes AI sector by limiting when labs can be held legally responsible, even for outcomes their systems might cause that are classified as "critical harm." The proposed legislation, which has sparked significant debate, would create a higher legal threshold for holding AI companies accountable. The term "critical harm" within the bill's text is understood to encompass scenarios of extreme consequence, including mass casualties or catastrophic financial collapses directly enabled by an AI system. OpenAI's position is that without clear liability boundaries, the fear of endless litigation could stifle the development of advanced AI, pushing research underground or out of the United States entirely. This stance aligns with a broader industry effort to shape favorable regulatory frameworks as governments worldwide grapple with AI governance. Critics, including some consumer advocacy groups and legal scholars, have condemned the push for liability limits as a dangerous precedent that could allow powerful corporations to evade responsibility for the societal impacts of their products. They argue that granting broad immunity could remove a crucial incentive for companies to build rigorous safety and alignment measures into their AI systems from the outset. The debate in Illinois is being closely watched as a potential bellwether for similar legislative efforts in other states and at the federal level, highlighting the fundamental tension between accelerating technological progress and ensuring corporate accountability for its potentially grave consequences.