Multistakeholder regulatory organizations.

Senator McNerney's proposal to establish Multistakeholder Regulatory Organizations (MROs) creates a new framework for certifying artificial intelligence models and applications in California. Under the legislation, the Attorney General would designate private entities as MROs for three-year renewable terms to evaluate AI systems' compliance with safety standards and best practices.

The bill requires MRO applicants to submit detailed plans addressing risk mitigation across areas including cybersecurity, biological threats, and AI model autonomy. These plans must outline approaches for auditing AI systems, ensuring developer disclosure of incidents, specifying certification parameters, and collecting evaluation data. MROs would be required to maintain independence from the AI industry through annual audits of their board composition, funding sources, and available resources.

Once designated, MROs would certify qualifying AI models and applications while decertifying those failing to meet requirements. The organizations must submit annual reports to the Legislature and Attorney General detailing AI capabilities, observed and potential risks, certification assessments, and prescribed remedial measures. The bill also establishes a rebuttable presumption in civil cases that developers exercised reasonable care if their AI system was MRO-certified when alleged injuries occurred. The Attorney General retains authority to revoke MRO designations for issues like misleading plans, compromised independence, or significant harms caused by certified systems.