OpenAI and Anthropic are backing competing bills in the Illinois General Assembly that would shape how frontier AI developers are held accountable when AI systems cause catastrophic harm. The bills focus on whether developers should face broad liability carveouts versus new public safety and child protection planning requirements. OpenAI supports SB 3444, which would shield frontier AI developers from liability for death or serious injury to 100 or more people, or property damage above $1 billion, with additional carveouts involving chemical, biological, radiological, or nuclear weapons. Anthropic opposes that approach, arguing it creates a “get-out-of-jail-free card” that weakens public safety and accountability. Anthropic instead supports SB 3261, which requires developers to publish a public safety and child protection plan, adds an incident reporting system for catastrophic risk events, and expands potential liability where models cause severe emotional distress, death, bodily injury, or self-harm for children. Experts cited in the piece say SB 3444 is unlikely to pass due to weaker enforcement mechanics.