Guides AI developers and users in California on compliance with existing laws governing consumer protections, data protections, civil rights protections, competition laws, and for new AI-specific laws.
Analysis summaries, actor details, and coverage mappings were LLM-classified and may contain errors.
This is a legal advisory document that provides guidance on how existing California laws apply to AI, rather than creating new binding legal obligations. It uses predominantly advisory language ('should', 'must ensure', 'should be aware') and serves to inform entities about their existing legal obligations under various California statutes.
The document has good coverage of approximately 12-14 subdomains, with strong focus on discrimination and toxicity (1.1, 1.2, 1.3), privacy and security (2.1, 2.2), misinformation (3.1, 3.2), malicious actors (4.1, 4.2, 4.3), human-computer interaction (5.1), socioeconomic impacts (6.2, 6.3), and system safety failures (7.3, 7.4). Coverage is concentrated in consumer protection, civil rights, privacy, and AI safety domains.
The document governs AI use across multiple sectors with explicit coverage of Information (social media platforms, telecommunications), Finance and Insurance (credit decisions, insurance claims), Health Care and Social Assistance (medical diagnoses, health insurance), Educational Services (K-12 educational technology), and Public Administration (election systems, government use of AI). Additional sectors receive minimal coverage through examples of AI applications.
The document addresses multiple lifecycle stages with primary focus on deployment and operation/monitoring. It covers design considerations through civil rights and consumer protection requirements, data collection through privacy laws, model development through training data disclosure requirements, validation through testing requirements, deployment through various disclosure and authorization requirements, and ongoing monitoring through reporting and oversight mechanisms.
The document broadly refers to 'artificial intelligence' and 'AI systems' throughout, providing a definition that encompasses machine-based systems that make predictions, recommendations or decisions. It does not specifically distinguish between frontier AI, general purpose AI, or task-specific AI, nor does it mention foundation models, compute thresholds, or open-weight models. It does reference generative AI in the context of content generation and deepfakes.
California Attorney General's Office (AGO)
The document is explicitly issued by the California Attorney General's Office to provide guidance on AI compliance with California law.
California Attorney General's Office; courts; regulatory agencies referenced in various statutes
While the advisory itself does not create enforcement mechanisms, it references existing enforcement under California statutes. The Attorney General's Office has enforcement authority under many of the referenced laws (UCL, CCPA, etc.).
California Attorney General's Office; Consumer Financial Protection Bureau
The document references monitoring and oversight responsibilities under various California statutes, though the advisory itself does not establish new monitoring mechanisms.
consumers; entities that develop, sell, and use artificial intelligence; businesses; nonprofits; government
The advisory explicitly targets AI developers, businesses that use AI, and consumers, providing guidance on their rights and obligations under California law.
15 subdomains (9 Good, 6 Minimal)