This is a research prototype. The data and analyses are preliminary and not yet validated — we'd welcome your .
Official name: California SB 503 (Health Care Services: Artificial Intelligence 2025)
Requires developers and deployers to identify, mitigate, and monitor AI systems in healthcare for biased impacts. Obligates them to report compliance efforts annually. Instructs the department to publish these reports online. Effective from January 1, 2027.
Analysis summaries, actor details, and coverage mappings were LLM-classified and may contain errors.
This is a binding California state statute with mandatory obligations on developers and deployers, enforceable reporting requirements, and legal compliance mechanisms.
The document primarily addresses discrimination and bias risks (1.1, 1.3) in healthcare AI systems, with focus on unfair treatment and unequal performance across protected groups. It also touches on governance (6.5) through regulatory oversight mechanisms and system safety (7.3, 7.4) through requirements for monitoring and documentation.
The document exclusively governs AI use in the Health Care and Social Assistance sector, with detailed requirements for developers and deployers of AI systems used in clinical decisionmaking and healthcare resource allocation. No other sectors are regulated.
The document covers multiple lifecycle stages with primary focus on deployment and operational monitoring. It addresses design considerations through bias identification requirements, deployment through pre-deployment reporting obligations, and ongoing monitoring through annual deployer reporting and continuous bias mitigation requirements.
The document explicitly defines and covers AI systems used for clinical decisionmaking and healthcare resource allocation. It does not mention frontier AI, general purpose AI, foundation models, generative AI, predictive AI, open-weight models, or compute thresholds. The focus is on task-specific healthcare AI applications.
The document is enacted by the California state legislature as indicated by the opening phrase 'THE PEOPLE OF THE STATE OF CALIFORNIA DO ENACT AS FOLLOWS' and is codified as Section 1339.76 of the Health and Safety Code.
The department is designated to receive compliance reports from developers and deployers and to publish these reports publicly, serving as the primary oversight body.
The department monitors compliance through mandatory reporting mechanisms and public disclosure of compliance efforts, though it explicitly does not independently inspect or test AI systems.
The statute explicitly targets both developers (those who design, code, or produce AI systems) and deployers (health facilities, clinics, physician's offices that use AI systems) for clinical decisionmaking or healthcare resource allocation.
5 subdomains (2 Good, 3 Minimal)
