OpenAI has expressed its support for Illinois bill SB 3444. According to WIRED, the bill stipulates that developers would not be held liable if a frontier model causes significant harm, provided the developer did not act intentionally or recklessly and has published reports regarding safety and other metrics.
This support appears to stem from a strategic shift. OpenAI has previously maintained a defensive stance, opposing bills that would hold AI developers accountable. However, some have pointed out that SB 3444 represents a more extreme measure than bills the company has supported in the past. A spokesperson for the company explained that they prioritize moving toward clear national standards over a patchwork of state-by-state rules.
The bill defines a frontier model as one trained with a computing cost of $100 million or more. This definition applies to major companies such as OpenAI, Google, and Anthropic. Examples of harm cited in the bill include cases where AI is misused to create chemical weapons.
Meanwhile, Scott Wisor of the Secure AI project stated that the likelihood of the bill passing is low, as Illinois is typically proactive in technology regulation. He also noted that there is significant opposition to granting liability exemptions to AI companies.
Source: OpenAI backs Illinois bill that would limit when AI labs can be held liable（HN 447pt・324コメント）（HN Search (backfill), 2026-04-10）