California Mandates Independent Audits for AI Systems
New laws in California require third-party verification of AI models, creating a state-level check on automated decision-making and safety.
California Governor Gavin Newsom has signed two bills that establish a formal framework for independent third-party audits of artificial intelligence systems. The legislation, enacted this week, creates a structured way to evaluate whether AI models comply with state laws, a significant step as these technologies become more embedded in government services and critical infrastructure.
The move adds a new layer of oversight to California’s existing regulatory landscape, which already includes rules for frontier AI developers regarding safety disclosures and incident reporting. By requiring external verification, the state aims to ensure that AI systems operate with meaningful safeguards that protect the public interest without stifling innovation.
New rules define auditor standards
One of the signed measures, SB 813, creates a framework for independent verification organizations that can assess AI systems and models. The other, AB 1405, establishes a state registry for AI auditors and sets specific standards for their independence, transparency, and integrity. This ensures that the entities checking the technology are not merely hired hands but are held to rigorous ethical and professional benchmarks.
According to Newsom’s office, these laws address concerns raised by recent incidents involving AI failures. The governor stated that while artificial intelligence holds extraordinary promise, it must be developed and deployed with robust safeguards to protect citizens. The framework is designed to provide objective, external validation rather than relying solely on internal self-assessments by technology companies.
State action fills federal gap
This regulatory effort follows a similar trend in other states, such as Illinois, which recently signed the Artificial Intelligence Safety Measures Act. That law mandates annual independent third-party audits for major frontier AI developers. In a statement, Illinois Governor JB Pritzker noted that as AI systems become more powerful and the federal government remains unwilling to step in, states have a responsibility to protect their people from potential dangers.
Newsom has explicitly called on the federal government to pass nationwide regulations, arguing that the scale and potential consequences of AI technology demand a national response. He emphasized that while California has taken leading action to advance safety and accountability, the urgency of the moment requires sustained action from every level of government to match the pace of technological change.
Government expands own AI use
The new audit requirements come at a time when California is rapidly expanding its own use of artificial intelligence in public services. The state recently launched AskCA, an AI-powered assistant designed to help residents navigate government services. State agencies have also gained access to advanced tools from major providers, and employees are testing internally developed assistants to improve workflow efficiency.
As the state increases its reliance on these tools, it is simultaneously strengthening oversight of AI vendors seeking state contracts. Recent measures require these vendors to address risks including bias, civil rights violations, and the generation of harmful content. This dual approach of expanding adoption while tightening compliance aims to balance operational efficiency with public safety.
