Overview
- OpenAI, which voiced support in statements published Friday, backed Illinois SB 3444 to shield developers of advanced AI models from lawsuits over large-scale disasters if they were not intentional or reckless and posted required safety and transparency reports.
- Coverage applies only to so‑called frontier models, defined in the bill as systems trained with more than $100 million in compute costs, a bar that would likely include OpenAI, Google, Anthropic, xAI, and Meta.
- The bill’s “critical harms” threshold targets rare, severe events such as the death or serious injury of 100 or more people, at least $1 billion in property damage, or an AI system enabling the creation of chemical, biological, radiological, or nuclear weapons.
- An OpenAI policy lead testified in favor of a federal framework to avoid a state‑by‑state patchwork, and Quartz reports the Illinois measure would stop applying if Congress later enacts overlapping rules.
- Policy experts quoted by WIRED say the bill faces long odds in Illinois and point to polling that found 90 percent of respondents oppose exempting AI firms from liability, reflecting the state’s tougher stance on tech regulation.