Newsom signs industry-backed AI auditing laws in California
The governor establishes a new framework for outside safety evaluations, marking a shift from previous restrictive legislative attempts.
California Governor Gavin Newsom signed two new bills in September 2026 establishing the nation's first auditing framework for artificial intelligence safety. The legislation creates a structured process for outside groups to evaluate and audit AI programs, signaling a new regulatory approach in the world's primary AI hub.
The new laws focus on the implementation of safety evaluations conducted by third-party entities. Unlike previous legislative efforts, these bills received explicit endorsement from major industry leaders, including OpenAI and Anthropic. By focusing on auditing and evaluation rather than rigid mandates, the framework seeks to balance safety oversight with the operational needs of AI developers.
A Shift in Strategy
This legislative move follows a period of significant tension between Sacramento and the tech industry. In 2024, Governor Newsom vetoed SB 1047, a more stringent AI safety bill. At the time, the governor argued that the requirements of that bill were too rigid and posed a risk of stifling innovation within the state.
The current laws represent a pivot toward an industry-aligned model. While SB 1047 sought to impose stricter, more prescriptive mandates on developers, the newly signed legislation prioritizes a framework of transparency and external review. This transition suggests a preference for collaborative safety standards over the heavy-handed regulatory constraints that characterized earlier attempts.
Industry Implications
Because California serves as the global center for AI development, its regulatory decisions often function as a de facto national standard for the United States. The adoption of an auditing-based framework suggests that the industry may be moving toward a model of "co-regulation," where companies agree to external oversight in exchange for avoiding more restrictive government mandates.
This shift highlights the ongoing struggle to manage the existential risks associated with rapidly advancing AI while maintaining a competitive edge in the global market. By securing the support of companies like Anthropic and OpenAI, the state has avoided a protracted legal or political battle with the very entities it seeks to regulate.
Looking Ahead
As the auditing framework is implemented, the industry will be watching to see how these outside evaluations are conducted and whether they result in meaningful changes to AI deployment. The effectiveness of this approach remains to be seen, particularly regarding whether voluntary industry alignment provides sufficient protection against catastrophic risks.
Observers will now monitor how other states respond to California's model and whether this auditing-centric approach becomes the blueprint for federal AI legislation. For now, the focus remains on the transition from the restrictive goals of 2024 to the collaborative auditing standards of 2026.