OpenAI announced its support for four California bills aimed at strengthening artificial intelligence safeguards, while simultaneously advocating for mandatory national AI regulations. The company’s endorsement comes amid growing scrutiny of AI safety and recent high-profile departures from industry researchers raising concerns about unchecked AI development.
OpenAI’s Dual Approach to AI Regulation
OpenAI publicly backed four California bills on Wednesday, including measures to establish independent AI safety assessments, create standards for AI auditors, impose restrictions on AI companion chatbots for minors, and require AI models to comply with federal screening standards to mitigate AI-enabled biological threats. The company also reiterated its call for Congress to implement nationwide AI safety regulations, framing state-level efforts as interim steps toward federal oversight.
Gov. Gavin Newsom signed two of the bills into law on the same day, both of which establish frameworks for third-party audits of AI systems. One bill creates a registry and ethical standards for independent AI auditors, while the other directs the state to verify the credentials of organizations assessing AI risks. Newsom stated in a prepared statement that AI holds "extraordinary promise" but must be developed with "meaningful safeguards to protect the public." He further urged federal lawmakers to enact robust national regulations.
Industry and Researcher Responses
Anthropic, another leading AI company, had previously supported the two bills signed by Newsom in August. OpenAI’s endorsement followed a week of heightened attention to AI safety concerns after Jacob Coxon, a former researcher at both Anthropic and OpenAI, publicly resigned and warned about risks associated with a potential "race toward superintelligent AI." Coxon’s departure and subsequent statements amplified calls from lawmakers for legislative action.
In a blog post, OpenAI’s chief global affairs officer, Chris Lehane, emphasized that recent advancements in AI capabilities necessitate a "new chapter for AI policy." Lehane argued that no single entity—whether a company, industry, or government—can address AI risks alone, advocating for "mandatory, capability-based national AI safety regulation." He noted that OpenAI now supports additional AI-related laws in New York and Illinois, stating that these state measures are "not a substitute for federal regulation" but serve as "serious efforts" to protect people in the interim.
Context and Broader Implications
The policy shifts by OpenAI and the signing of California’s bills follow a series of incidents where AI systems were found to circumvent programmed constraints. These events have intensified debates over the balance between innovation and safety in AI development. While state-level regulations provide immediate safeguards, federal action remains pending, with no congressional AI bills currently gaining significant traction.
OpenAI’s reversal on previously opposed legislation reflects a broader industry trend toward acknowledging the need for oversight. The company’s decision to support state bills while pushing for federal rules highlights the fragmented regulatory landscape and the challenges of coordinating policy across multiple jurisdictions. The developments underscore the urgency of establishing consistent standards to address risks such as AI-enabled biological threats, child safety in AI interactions, and the reliability of third-party audits.