The bill mandates that the Government Operations Agency designate qualified independent verification organizations (IVOs) to conduct rigorous safety audits. While existing law requires frontier developers to publish safety frameworks and disclose critical incidents, these measures currently rely on companies evaluating their own systems. SB 813 introduces a layer of external accountability intended to protect businesses, families, and children from potential AI-driven risks.
California Advances Landmark AI Safety Verification Bill
California lawmakers passed SB 813 late last night, a move that would establish the nation’s first state-level framework for independent safety testing of artificial intelligence models. Authored by Senator Jerry McNerney and backed by the nonprofit Fathom, the legislation shifts risk assessment away from internal corporate reporting toward external oversight.
Fathom CEO Andrew Freedman stated that the model allows qualified experts to assess evolving AI capabilities without locking specific testing methods into rigid law. This flexibility aims to keep pace with the rapid advancement of the technology. If Governor Gavin Newsom signs the bill, California will become the first state to build the necessary infrastructure to support a formal ecosystem of third-party evaluators, setting a precedent that other states and federal policymakers may soon follow.
Comments (0)
No comments yet. Be the first!