California Establishes Framework for Independent AI Assessors

California Governor Gavin Newsom recently signed two bills establishing a framework for independent AI assessments and creating a state registry for AI auditors. These measures aim to increase transparency and credibility in outside reviews of increasingly capable AI systems as they become integrated into everyday products and critical services.

Senate Bill 813, authored by state Senator Jerry McNerney, and Assembly Bill 1405, authored by Assemblymember Rebecca Bauer-Kahan, outline the framework and registry requirements. The laws establish a role for independent organizations to verify AI systems and models, emphasizing the need for outside assessors to meet common standards for independence, transparency, and integrity.

The framework intends to help evaluate whether external assessors meet these standards. The move comes as AI systems become more powerful and interconnected, making independent verification crucial to prevent potential risks from affecting broader economic and public life. While registration won't replace an organization's own security testing and monitoring, it could serve as an additional due-diligence resource for IT and procurement leaders evaluating AI vendors or commissioning outside reviews.

California's framework adds to its growing set of AI regulations, previously including disclosure requirements for frontier AI developers and measures addressing AI-generated content, privacy, fraud, and cybersecurity. The state also launched an AI Cyber Defense Program in August and is tightening rules around AI used by children.

The bills aim to provide a structured approach for organizations to assess AI systems, potentially impacting costs and timelines for deployments, especially for smaller companies with limited compliance resources. However, the framework also addresses the challenge of rapidly advancing AI capabilities outpacing traditional oversight systems.