OpenAI's disavowal of a liability shield in Illinois SB 3444 bill and endorsement of a stronger SB 315 suggest it is open to meaningful AI safety legislation
Transformer Weekly: US-China talks, AI executive order, and Anthropic's $900b valuation … - Scott Bessent said the US and China will …
Context & Ripple Effects
Related coverage first described OpenAI backing Illinois SB 3444, whose reporting-based safe harbor could have limited laboratory liability even after severe harms. Anthropic publicly opposed that approach, exposing a split between major AI developers over whether transparency alone should constrain liability.
OpenAI has since endorsed SB 315, which the company describes as adding transparency and incident-reporting requirements, while disavowing the SB 3444 liability shield. That is a meaningful shift from arguing that a California safety bill should be handled federally rather than through state legislation.
First-order effects
- OpenAI’s Illinois policy position now favors a bill with affirmative safety obligations over a broad liability protection tied to publishing safety reports.
- SB 3444 loses a prominent supporter for its contested safe-harbor provision, while SB 315 gains backing from a leading AI lab.
Second-order effects
- Illinois lawmakers can treat the change as evidence that disclosure and incident-reporting rules may be more commercially and politically viable than sweeping immunity for model developers.
- Other AI labs and industry groups face greater pressure to state whether they support enforceable state-level safety duties, rather than relying primarily on voluntary reports or federal preemption arguments.
Third-order effects
- If this stance persists, AI regulation may develop around baseline operational requirements—transparency, incident reporting, and defined accountability—rather than blanket protections for frontier-model providers.
- The episode also suggests state bills can become testing grounds for AI-governance compromises, though whether they converge into durable, interoperable rules or a patchwork remains unresolved.
The trend: Major AI developers are moving from broad resistance to state safety legislation toward selective support for rules that formalize transparency and incident accountability without adopting expansive liability shields.