California does not yet have a single, fully enacted statute or amended Rule of Professional Conduct that is uniquely dedicated to attorney use of AI. As of September 2, 2026, the governing framework is a combination of existing ethics rules, detailed State Bar guidance, court rules for the judiciary, and pending legislation and rule amendments.
Since ABA Formal Opinion 512 gave lawyers nationwide their first real ethical roadmap for generative AI in July 2024, California has moved further and faster than any other state, shifting from advisory guidance toward binding rules with real teeth. Below is where the rules actually stand today, what’s still just proposed, and other relevant information.
Attorneys must comply with the existing California Rules of Professional Conduct and the State Bar Act when using any technology, including generative or agentic AI. The most relevant existing rules are:
– – Rule 1.1 (Competence) — Includes keeping abreast of relevant technology. A lawyer remains responsible for the work product and cannot delegate professional judgment to AI. Outputs must be reviewed and verified.
– – Rule 1.6 and Bus. & Prof. Code § 6068(e) (Confidentiality) — Do not input confidential client information into a public or inadequately secured generative AI system. “Reveal” can include exposing information to an AI tool that trains on inputs or lacks adequate security.
– – Rule 3.3 (Candor toward the tribunal) — A lawyer must not make false statements of law or fact. Courts have sanctioned lawyers for filing briefs with AI-hallucinated citations that the lawyer did not personally read and verify.
– – Rules 5.1 and 5.3 (Supervision) — Firm leaders must have policies on AI use; lawyers must supervise nonlawyer assistants (and, by extension, AI tools) so that ethical duties are met.
– – Rule 1.4 (Communication) — Clients must be kept reasonably informed. Material use of AI that affects cost, risk, or the manner of representation may need to be discussed.
– – Rule 1.5 (Fees) — You may charge for time spent prompting, reviewing, and editing AI output. You generally should not bill as if you performed the work the AI performed without adjustment for efficiency.
These duties already apply; the State Bar’s guidance simply maps them onto AI.
Another binding rule is California Rule of Court 10.430, effective September 1, 2025 — but it’s worth being precise about who it actually governs. Rule 10.430 requires every California court that permits generative AI use to adopt a written AI use policy by December 15, 2025, covering court staff for any purpose and judicial officers for any task outside their adjudicative role (a companion provision, Standard of Judicial Administration 10.80, separately covers judges’ adjudicative use). It is a rule for the courts themselves, not a rule directly regulating how outside attorneys or law firms use AI in their own practices.
The Standing Committee on Professional Responsibility and Conduct (COPRAC) issued Practical Guidance for the Use of Generative Artificial Intelligence in the Practice of Law in November 2023 and replaced it with an updated version approved by the Board of Trustees in May 2026. The update specifically addresses agentic AI (systems that can plan and execute tasks with limited human prompting). It is guidance, not a disciplinary rule, but it is the most detailed official statement and is the document the California Supreme Court directed the Bar to consider incorporating into the Rules.
Core principles from the 2026 Guidance:
– – Do not input confidential or nonpublic client information into a public generative AI system that lacks adequate security and confidentiality protections.
– – Attain a reasonable understanding of the specific tool’s capabilities, limitations, data sources, and risks before using it.
– – Independently review, verify, and exercise professional judgment over every output used in a representation. Detecting hallucinations is necessary but not sufficient.
– – Do not let agentic systems make substantive legal determinations, give legal advice, or file pleadings without meaningful lawyer supervision and review.
– – The lawyer remains fully responsible for all work product.
– – Take reasonable steps to avoid biased or discriminatory outputs.
– – Fees should reflect actual time and value; AI efficiency should not be billed as if the lawyer performed the work unaided.
– – Comply with all other applicable law (privacy, IP, cybersecurity, AI-specific statutes, etc.).
The rule that will actually bind practicing attorneys directly is still in progress. On direct order from the California Supreme Court, the State Bar’s Committee on Professional Responsibility and Conduct (COPRAC) has drafted proposed amendments to six Rules of Professional Conduct addressing AI. In response to an August 22, 2025 letter from the California Supreme Court, COPRAC proposed comments to Rules 1.1, 1.4, 1.6, 3.3, 5.1, and 5.3. A second public-comment period ran after the June 12, 2026 COPRAC meeting (deadline August 6, 2026). The proposals still require Board of Trustees and Supreme Court approval and are not yet binding. If adopted they would, among other things:
– – Explicitly require independent review and verification of AI (and other technology) outputs.
– – Clarify that “reveal” under Rule 1.6 includes exposing confidential information to AI systems that create a substantial risk of inconsistent use.
– – Require verification that cited authorities exist and are accurate before submission to a tribunal (including AI-assisted citations).
– – Require managerial lawyers to establish internal AI policies and to instruct/supervise nonlawyer assistants on ethical use of AI.
Legislation is moving on a parallel, faster track. Senate Bill 574, introduced by Senate Judiciary Chair Tom Umberg passed both houses unanimously and was ordered to engrossing and enrolling on August 31, 2026. As of September 2, 2026 it has been sent to Governor Newsom but has not yet been signed. The Governor typically has until September 30 to act on end-of-session bills.
If enacted, SB 574 would add a new Business and Professions Code section 6068.1 codifying attorneys’ AI duties (protecting confidentiality, verifying accuracy, avoiding bias, considering disclosure for public-facing content), and would amend the state’s existing sanctions statute, Code of Civil Procedure section 128.7, to make it explicit grounds for sanctions if a court filing contains any citation the responsible attorney has not personally read and verified — whether or not AI produced it.
Separately, the bill would add Code of Civil Procedure section 1282.1, barring arbitrators from delegating any part of their decision-making to generative AI, barring reliance on AI-sourced information outside the arbitration record without disclosure, and holding the arbitrator personally responsible for the award regardless of AI assistance used along the way.
The landscape is moving quickly. The Practical Guidance is the best current official map; the proposed rule comments and SB 574, if they become law, will convert much of that guidance into enforceable duties. Check the State Bar’s Ethics & Technology page and the official bill status for updates.