California Government AI Accountability Act (SB 896)
Effective date
Penalty
Government accountability. No private sector penalties.
Obligations mapped
Tracked
Overview
Requires California state agencies to disclose use of generative AI in communications with individuals about government services and benefits. Requires OES to conduct annual analysis of risks to state critical infrastructure from generative AI. Applies to government entities, not private sector.
This is an AI-specific state law.
See if this regulation applies to your company with the free exposure scan.
Who this applies to
This regulation applies to the following roles:
- Developers of covered AI systems
- Deployers and users of covered AI systems
- Organizations operating in California
This regulation applies to both companies that build AI products and companies that use AI tools from other vendors.
SB 896
AI categories covered
- Government AI use
- General purpose AI
What this requires you to do
Detailed obligation packs are not yet mapped for this entry in XIRA. Obligation areas from the catalog are listed below.
What this requires you to do
Disclose AI use. Make it clear to users when they are interacting with AI-generated content or AI-driven systems.
Provide transparency notices. Inform affected individuals that AI is being used and how it influences decisions.
Implement a risk management program. Maintain ongoing processes to identify, assess, and mitigate AI-related risks.
Maintain records. Keep documentation of your AI systems, decisions made, and compliance activities.
Regulation summaries are simplified for readability and may not capture every nuance of the underlying statute. Verify important details against primary sources linked on this page.
Enforcement and penalties
Government accountability. No private sector penalties.
Penalty amounts are based on statutory text and may be subject to adjustment, judicial interpretation, or enforcement discretion.
Legislative history
effective
Takes effect
Related regulations
- In EffectAI-Specific
Texas Government AI Ethics and Oversight (SB 1964)
Requires Texas state agencies and local governments to inventory AI systems, adopt an AI code of ethics aligned with NIST AI RMF, conduct impact assessments for AI that autonomously influences consequential decisions, and disclose AI use to affected individuals. Applies to government entities only, not the private sector.
Effective
- UpcomingAI-Specific
California AI Transparency Act (SB 942)
The California AI Transparency Act requires creators of large generative AI systems to provide free AI detection tools, embed provenance metadata in AI-generated content, and offer visible disclosure options. Large platforms must detect and preserve provenance data.
Effective
- UpcomingPrivacy ADM
CCPA/CPRA Automated Decision-Making Technology Regulations
California's ADMT regulations require businesses using automated decisionmaking technology for significant decisions (employment, finance, housing, education, healthcare) to provide pre-use notices, offer opt-out rights, respond to access requests, and conduct risk assessments with annual CPPA filing under penalty of perjury.
Effective
- In EffectAI-Specific
California Transparency in Frontier AI Act (SB 53)
Requires developers of frontier AI models trained above the statutory compute threshold (10^26 FLOPs) to publish safety frameworks, report critical safety incidents to the Office of Emergency Services, and implement whistleblower protections. Also reaches large frontier developers with annual revenues over $500 million. Replaces the vetoed SB 1047 with a narrower transparency approach. Currently applies to approximately five to eight companies worldwide given the FLOP threshold. Includes a federal deference provision: compliance with comparable federal standards, including the EU AI Act, is accepted where the statute allows.
Effective
- In EffectAI-Specific
California AI Training Data Transparency Act (AB 2013)
Requires developers of generative AI systems or services available to Californians to publish high-level documentation on training data, including sources, types, and curation. Applies retroactively to systems released on or after January 1, 2022. No trade secret exemption. Internal development or material modification of third-party GenAI can be in scope. xAI challenged the law in federal court; on March 4, 2026 the court denied a preliminary injunction, so AB 2013 remains in full effect while litigation continues. Major providers published required documentation by January 1, 2026.
Effective
- In EffectAI-Specific
California Healthcare Provider Generative AI Disclosure (AB 3030)
Requires healthcare providers to disclose when generative AI is used in patient interactions and to document that use in the patient record. Focuses on licensed providers and clinical settings. AB 489 (2025) later extended parallel transparency duties to developers and deployers of healthcare AI, not only providers.
Effective
- In EffectAI-Specific
California Healthcare AI Deceptive Terms Act (AB 489)
AB 3030 (2024) requires healthcare providers to disclose generative AI use to patients and in records. AB 489 (2025) extends similar duties to technology developers and deployers whose healthcare AI communicates with patients or presents as credentialed care. It bars false claims of professional licenses or credentials and requires clear disclosures in healthcare settings.
Effective
- In EffectAI-Specific
California FEHA regulations on automated decision systems (Civil Rights Council)
California Civil Rights Council regulations apply FEHA's anti-discrimination framework to automated decision systems (ADS) in employment. Defines ADS broadly to include AI, ML, and algorithmic tools. Makes anti-bias testing evidence relevant to discrimination claims and defenses. Requires reasonable accommodation when ADS disadvantages disabled or religious individuals. Prohibits pre-offer medical inquiries via ADS. Employers with 5+ employees are covered. 4-year record retention required.
Effective
- In EffectAI-Specific
California Companion Chatbots Act (SB 243)
California's Companion Chatbot Act may apply to operators of AI chatbots designed for ongoing social interaction. Where applicable, operators may need to disclose the AI nature of the chatbot, maintain safety protocols for self-harm and suicide content, provide crisis referrals, and implement special protections for minors including break reminders and content restrictions. Operators may need to publish safety protocols and file annual reports with the Office of Suicide Prevention starting July 2027.
Effective
- In EffectAI-Specific
California Digital Replicas of Deceased Performers Act (AB 1836)
Restricts commercial uses of realistic AI-generated replicas of deceased performers' voices or likenesses in audiovisual works and sound recordings without consent from the personality's estate or other rightsholder.
Effective
- In EffectAI-Specific
California Digital Replica Contract Protections (AB 2602)
Makes contract terms unenforceable when they allow digital replicas of a performer without the performer giving specific, informed consent. Two paths to enforceable consent: independent legal counsel review or union agreement. Companion to AB 1836.
Effective
- In EffectAI-Specific
California AI Definition Act (AB 2885)
Standardizes the legal definition of artificial intelligence across all California law. Defines AI as an engineered or machine-based system that varies in its level of autonomy and can infer from input how to generate outputs that influence physical or virtual environments. Mirrors the OECD AI definition.
Effective
California AI regulation guide lists every tracked rule for this jurisdiction with timelines and obligation tallies.
Regulation summaries are simplified for readability and may not capture every nuance of the underlying statute. Verify important details against primary sources linked on this page.