OpenAI on Wednesday testified before Illinois legislators in favor of a bill that would create a liability shield for AI companies, activating the shield only when a product causes at least 100 deaths.

The testimony drew attention because OpenAI is simultaneously defending against wrongful death lawsuits filed by families of teenagers who died by suicide, according to Forbes.

The juxtaposition has intensified debate over how the United States should assign legal accountability for harms caused by AI systems.

What The Illinois Bill Proposes

The bill, as described in the Forbes report, would treat AI-related liability under a "catastrophic harm" framework. A triggering threshold of 100 deaths would need to be met before the liability protection falls away. Below that number, companies would retain substantial insulation from civil claims.

Proponents of the approach argue it mirrors how other industrial liability regimes work, focusing resources on systemic failures rather than individual incidents. Critics argue the threshold is calibrated to protect companies rather than victims.

The bill has not passed. It remains in committee deliberation as of May 27.

OpenAI's Legal Position

OpenAI faces multiple wrongful death suits tied to interactions with its AI products. In at least some cases, families of teenagers allege that conversations with AI systems contributed to mental health deterioration and eventual suicide. The company has disputed the causal claims. It has argued that the appropriate standard for liability should account for the scale and unpredictability of AI deployment.

The Illinois testimony asking for a 100-death threshold was made while those suits remain active. That timeline has drawn criticism from plaintiff attorneys and from some legislators. An AI company simultaneously defending individual harm cases while lobbying for high collective-harm thresholds presents an unusual legal posture.

Background

The question of AI liability has moved rapidly from academic debate into active legislation in 2026. California's Governor Newsom signed an executive order this week directing state agencies to study AI workforce displacement, a separate policy track but part of the same broader pressure on governments to respond to AI-related harms. Colorado rewrote its AI accountability law before it took effect, reducing liability provisions significantly.

The federal government has not passed comprehensive AI liability legislation. That gap has pushed the fight into state capitols.

OpenAI reached a valuation of $730 billion in its most recent funding round. The scale of the company's commercial ambitions has sharpened scrutiny of its policy positions. Lobbying for minimal liability thresholds at that valuation level draws a different kind of attention than it might for a smaller startup.

Also Read: Anthropic Moves Restricted Claude Mythos Model Closer To Public Release

The Liability Design Problem

Fortune published a competing perspective this week arguing that pre-deployment safety testing cannot predict real-world AI behavior. The piece suggests liability frameworks anchored to measurable deaths or injuries may miss more diffuse harms, including economic displacement, mental health impact, and erosion of privacy.

That framing would imply the Illinois threshold approach is doubly flawed: too high a bar for catastrophic events, and the wrong metric entirely for chronic harms.

The tension between these two frameworks, industry-friendly thresholds versus harm-category coverage, is the central design problem facing US AI liability law. No state has resolved it. Illinois may become the first to codify one side of the argument.

What Comes Next

The Illinois bill faces substantial amendment pressure before any vote. Legal scholars have noted the 100-death threshold would likely face constitutional challenges if passed, given due-process concerns about foreclosing individual tort claims. OpenAI's lobbying position may shift as the wrongful death suits progress toward discovery.

Both the California executive order and the Illinois bill are expected to generate federal legislative proposals in the second half of 2026.

The outcome in Illinois will be watched closely by AI companies operating at scale. A precedent establishing high liability thresholds in one state creates pressure on others to adopt similar frameworks, or reject them explicitly.

Read Next: SUI Must Reclaim $1.31 To Avoid A Deeper Structural Breakdown