California AI Transparency and Governance

Overview

California SB 1159, chaptered as Chapter 863, Statutes of 2026, amends multiple transparency, administrative, open-meeting, records, and environmental statutes to address how artificial intelligence may interact with government processes. The measure clarifies that, for purposes of the California Public Records Act, the Legislative Open Records Act, the Bagley-Keene Open Meeting Act, the Ralph M. Brown Act, the Administrative Procedure Act, the California Coastal Act of 1976, and CEQA, terms such as “person,” “interested person,” “participant,” and “member of the public” do not include artificial intelligence systems, autonomous agents, or robots, whether physical or digital. The act also authorizes specified public bodies to use a disclosure verification tool meeting criteria in Business and Professions Code Section 22757.2 to determine whether AI is present in an engagement with government.

The statute is framed as a transparency and governance response to machine-generated mass participation. The Legislature states that existing access laws are designed to enable natural persons to participate in and observe governmental processes, while AI can automate interactions at scale and speed beyond human capacity, potentially flooding agencies with requests, comments, petitions, or other submissions. The act also adds a separate prohibition in Government Code Section 8319 against knowingly using AI to falsely represent that a natural person appeared before, submitted information to, or otherwise engaged with a governmental agency. The law preserves ordinary human use of AI as an assistive tool, including assistive technologies for people with disabilities, so long as the volume and frequency of the engagement remain reasonably consistent with ordinary participation by a natural person. The statute was approved and filed on 2026-09-30 and takes effect on 2027-01-01.

Definitions

SB 1159 uses a repeated statutory definition of “artificial intelligence” across the affected sections. AI is defined as an engineered or machine-based system that varies in its level of autonomy and that can, for explicit or implicit objectives, infer from the input it receives how to generate outputs that can influence physical or virtual environments. The act also repeatedly defines “assistive technology” as an item, piece of equipment, or product system, whether commercially acquired, modified, or customized, used to increase, maintain, or improve the functional capabilities of individuals with disabilities, plus any service that directly assists an individual with a disability in selection, acquisition, or use of the item or system. These definitions are important because they limit the scope of the exclusions and preserve disability-related accessibility use cases.

The operative drafting approach is chapter-specific rather than global. Each amended or added section states that the relevant governmental participation terms do not include AI systems, autonomous agents, or robots. In several sections, the statute then adds a parallel clause confirming that a natural person may use AI, autonomous agents, or robots to facilitate the person’s own engagement with government, including through assistive technologies. This structure draws a sharp line between machine-authored, machine-scaled participation and human participation assisted by technology. The law also gives government entities flexibility to use a disclosure verification tool that meets the criteria described in Business and Professions Code Section 22757.2, but it does not itself define that tool in the body of SB 1159.

Governance and Institutional Framework

The act operates across multiple institutions that manage California’s democratic and administrative processes. It reaches state agencies, the Legislature, the Office of Administrative Hearings, local agencies, and environmental and coastal authorities. In the APA-related provisions, “agency” and related participation terms are narrowed so AI systems do not count as persons or participants in regulatory rulemaking and related administrative processes. In the legislative records provision, the appropriate Rules Committee of each house and the Joint Rules Committee are expressly authorized to use a disclosure verification tool. In the Bagley-Keene Open Meeting Act section, a state body may use the same kind of tool. In the local government open-meeting provision, a local agency may do likewise. In the administrative adjudication provision, the Office of Administrative Hearings may use the tool.

The governance model is therefore decentralized but consistent: each public body retains authority to decide whether its proceedings or submissions involve AI, while using a standardized verification concept tied to another California law. The statute does not create a new oversight agency, licensing regime, or reporting office. Instead, it redefines who counts as a participant in existing public-law processes and gives agencies a procedural tool to manage that determination. The legislative findings in Section 13 expressly link the act to the constitutional right of access and say the clarification serves the public interest by preserving democratic integrity, preventing automated systems from displacing genuine human participation, and protecting public resources from machine-generated requests.

Key Focus Areas

The most prominent policy focus is transparency and disclosure. SB 1159 responds to concerns that AI could be used to submit mass comments, petitions, public records requests, or other communications that appear human but are not. By excluding AI from participation terms, the act aims to preserve the meaning of public comment and public access laws as channels for natural persons and legally recognized entities acting through human decision-making. A second focus is governance integrity. The statute treats government forums as deliberative spaces whose legitimacy depends on actual human judgment, and it specifically prohibits knowingly misrepresenting AI-generated engagement as human participation before a governmental agency.

A third focus is accessibility and accommodation. The act repeatedly protects human use of AI as a facilitative tool, including assistive technologies for individuals with disabilities. The statute does not ban AI-assisted drafting, submission, or participation by a human being; instead it targets engagement volumes and patterns that exceed ordinary human participation or that are used to impersonate human activity. A fourth focus is institutional resilience. The Legislature states that government bodies should not be overwhelmed by automated submissions that consume limited public resources. Taken together, the key policy objective is not AI prohibition, but preservation of authentic civic participation and administrative functionality in public records, rulemaking, open meetings, legislative records, adjudication, and environmental review processes.

Implementation Framework

The implementation model is statutory and immediate upon effectiveness, but many practical duties are embedded in the text of the amended sections rather than in a separate implementation chapter. Each affected entity may use a disclosure verification tool that meets the criteria in Business and Professions Code Section 22757.2 to determine whether AI is present. That tool authorization applies to a public agency under the Public Records Act, the Legislature through its Rules Committees, a state body under Bagley-Keene, the Office of Administrative Hearings, an agency under the APA chapter amendments, a local agency under the Brown Act, and a public agency under CEQA and the Coastal Act amendments. The act does not require adoption of the tool; it permits use.

Implementation also requires operational distinctions between prohibited conduct and permitted assistive use. Public bodies must not treat a natural person’s use of AI as disqualifying if the engagement volume and frequency are reasonably consistent with ordinary participation by a natural person. Conversely, a person may not knowingly use AI to falsely present human participation. Agencies therefore must calibrate review processes to detect synthetic engagement without deterring legitimate accessibility tools or human-assisted submissions. The statute’s practical effect is likely to require internal guidance, staff training, intake screening practices, and records-handling protocols, but those measures are not expressly mandated by name in the chapter. Because the act amends existing government codes rather than establishing a separate program, its implementation is distributed across existing public-law systems.

Monitoring and Evaluation

SB 1159 includes no stand-alone audit regime, annual report, or formal performance review mechanism. Instead, monitoring is embedded in the ability of public bodies to use a disclosure verification tool to determine whether AI is present. That authorization is the principal compliance-monitoring mechanism. The statute does not prescribe technical standards for how the tool must operate beyond requiring that it meet the criteria in Business and Professions Code Section 22757.2. Nor does it require agencies to publish detection rates, false positive rates, or usage statistics. As a result, evaluation is left largely to ordinary administrative management, legal challenge, and the practical consequences of agency review.

The legislative findings suggest the evaluative yardstick: whether the clarification preserves the integrity and functionality of democratic institutions, prevents automated systems from displacing genuine human participation, protects public resources from machine-generated requests, and ensures governmental decisions remain responsive to the people of California. In this sense, monitoring is not just about identifying AI; it is about preserving the quality of participation in public processes. Agencies likely must compare the volume and frequency of submissions against what is reasonably consistent with ordinary human participation when deciding whether a submission falls within the preserved human-use exception. The statute does not assign a separate enforcement authority for ongoing evaluation, leaving oversight distributed among the affected agencies and, ultimately, courts if disputes arise.

Penalties, Liability, and Appeals

The act creates a direct prohibition in Government Code Section 8319: a person shall not knowingly use artificial intelligence to falsely represent that a natural person appeared before, submitted information to, or otherwise engaged with a governmental agency. The text does not specify a dedicated civil penalty, criminal punishment, or administrative fine in the chapter itself. It also does not create a bespoke private right of action or damages scheme. Any consequence for violations would therefore depend on the interaction of this new prohibition with other applicable law, agency procedures, or enforcement mechanisms outside SB 1159.

Similarly, the measure does not establish a special appeals process for an agency’s determination that AI was present. Because the law operates through existing public-law systems, any challenge would likely proceed under the relevant underlying statute or general administrative law principles, depending on the context: public records, legislative records, open meetings, APA proceedings, coastal decisions, or CEQA review. The act’s preservation clause for human-assisted use is particularly relevant in disputes, because it limits the reach of the prohibition to engagements that are not reasonably consistent with ordinary participation by a natural person. That means enforcement questions will likely turn on whether the communication was synthetic impersonation, whether the user knowingly misrepresented AI-generated engagement, and whether the submission was actually a human’s own participation facilitated by AI.

Relationship to Other Instruments

SB 1159 directly amends or adds provisions to a set of long-standing California transparency and administrative statutes. Those include the California Public Records Act, the Legislative Open Records Act, the Bagley-Keene Open Meeting Act, the Ralph M. Brown Act, the Administrative Procedure Act, the California Coastal Act of 1976, and CEQA. It is therefore best understood as a cross-cutting clarification law that overlays existing access and participation rules rather than replacing them. The bill also references Business and Professions Code Section 22757.2 as the standard for disclosure verification tools, which makes that separate provision functionally important to the act’s operation.

The legislative findings expressly state that the act’s changes to the Public Records Act and the Brown Act further the purposes of Article I, Section 3 of the California Constitution concerning public access to meetings and writings of local public bodies and local agencies. The findings also state that the changes should not be construed to imply broader interpretations of “person” or related terms in laws outside the referenced transparency and governance statutes. That disclaimer matters because it keeps SB 1159 tightly cabined. It is not a general AI personhood statute, not a universal definition of AI across California law, and not a general rule for every state or local program. It is a targeted governance instrument for a defined family of public-law processes.

National/Federal Alignment

SB 1159 reflects a state-level approach that is broadly consistent with federal AI governance themes focused on authenticity, accountability, and protection against automation-driven abuse, while remaining structurally distinct from federal law. The statute’s core move is to preserve genuine human participation in democratic processes and prevent AI systems from masquerading as people in government forums. That orientation aligns with general federal concerns about transparency and identity in digital government, but SB 1159 is not implementing a federal mandate. It is a California-specific legal rule operating under state constitutional and statutory authority.

The statute itself references the U.S. Patent and Trademark Office’s November 2025 guidance and the European Union’s AI Act as contextual support in the legislative findings, but those references do not make California law subordinate to federal or EU standards. Instead, they underscore the Legislature’s view that AI is a tool for human use, not an independent civic actor. No federal AI statute cited in the text preempts or displaces this measure, and nothing in the chapter indicates a conflict with federal administrative, privacy, or records law. The act therefore represents complementary state governance: it addresses AI-generated participation risks in California’s own public-law systems, leaving federal law to govern federal agencies and federal processes separately.

Implementation Timeline

Compliance Checklist

Sources and References