California Just Made AI Disclosure Mandatory in Every Court Filing — Effective Immediately

California became the first US state to require mandatory AI disclosure in every court filing — a sweeping rule adopted by the Judicial Council this morning that takes effect immediately and reshapes the workflow of every litigator practicing in the state.

On the morning of May 13, 2026, the California Judicial Council voted 19–2 to adopt new Rule of Court 2.150 — the most aggressive AI disclosure mandate yet adopted by any US court system. Effective immediately, every lawyer filing any document in any California state court must disclose whether generative AI was used in its preparation, identify which tool was used, and certify that a human attorney verified every citation and factual assertion. There is no grace period. There are no exemptions for solo practitioners, no carve-outs for in-house counsel, and no transition relief for big-firm litigators.

What Rule 2.150 Actually Requires

The rule applies to every brief, motion, declaration, discovery response, and proposed order filed in any California superior court, court of appeal, or the California Supreme Court. Each filing must include a one-paragraph AI Use Disclosure on the signature page identifying: (a) whether generative AI was used at all; (b) the specific tool or platform; (c) the categories of work performed by the AI (research, drafting, summarisation, citation generation); and (d) an attorney certification that all citations have been independently verified against the underlying authority.

The Five Specific Obligations

• Disclose every AI tool used in preparing the filing — including Westlaw Precision AI, Lexis+ AI, Harvey, ChatGPT, Claude, Gemini, and any in-house deployment.
• Verify every citation by reading the source — the rule explicitly bars citations confirmed only by an AI tool.
• Identify the categories of AI work performed: legal research, drafting, summarisation, translation, or analytical reasoning.
• Sign a certification on the signature page — false certifications carry State Bar referral as a stated consequence.
• Preserve the AI session log for two years and produce it on court order in any matter where AI accuracy becomes a disputed issue.

Why California, and Why Now

The rule was driven by a 14-month working group convened after a string of AI-hallucination sanctions cases — most notably the December 2025 Sacramento County matter in which a litigation partner was suspended for 90 days after filing a brief containing six fabricated case citations generated by an unsupervised consumer AI tool. The Judicial Council's report cited 41 documented hallucination incidents in California state courts in 2025 alone, up from 9 the prior year. Chief Justice Guerrero used the announcement to deliver the line every California litigator will be reading tomorrow morning: the duty of candor to the tribunal is not delegable to a machine.

Sanctions Are Real, Specific, and Tiered

Rule 2.150 establishes a three-tier sanction structure. Tier one — failure to disclose — triggers monetary sanctions up to $10,000 per filing. Tier two — disclosure with unverified citations that turn out to be fabricated — triggers mandatory State Bar referral. Tier three — knowing false certification — is treated as a violation of California Rules of Professional Conduct 3.3 (candor to the tribunal) and 8.4 (misconduct), with disbarment squarely on the table. Judges have discretion to impose sanctions sua sponte; opposing counsel may also raise the issue by motion.

Who Is Covered

• Every attorney admitted in California, including out-of-state lawyers admitted pro hac vice.
• In-house counsel filing in their employer's name — no corporate exception.
• Self-represented litigants who use AI in preparing their own filings (with a simplified disclosure form).
• Federal-court practitioners whose state-court filings touch a parallel proceeding.

If you cannot tell the court which tool wrote the brief and which lawyer read every case it cited, you should not be filing the brief.

The Vendor Reaction

Within four hours of the announcement, the major legal-research vendors had updated their product pages with California-specific compliance language. Thomson Reuters confirmed that Westlaw Precision AI session logs are exportable in the Rule 2.150 format. LexisNexis published a one-click 'California Disclosure Generator' inside Lexis+ AI. Harvey announced a Rule 2.150 audit trail feature for all California matters. Casetext, Robin AI, and Spellbook issued similar updates by lunchtime. The vendors that did not respond — including several consumer-grade AI tools popular with solo practitioners — are now the highest-risk choice on the market.

What Litigators Should Do Today

• Update every brief template and motion template in your document management system to include the Rule 2.150 disclosure block.
• Audit which AI tools your team is actually using — including the unsanctioned ones — and approve a short list with audit-trail capability.
• Train every associate, paralegal, and contract attorney this week on the verification requirement: every cited case must be opened and read.
• Update your conflicts and intake software to flag California state-court matters for Rule 2.150 workflow.
• Brief your malpractice carrier — at least two major insurers have signalled that uninsured Rule 2.150 sanctions may not be covered.

What This Means Outside California

California historically sets the pace for state-court technology rules — the Texas, Florida, and New York judicial bodies all have AI working groups that have been waiting for a model rule to follow. Expect copycat rules in at least five additional states before year-end. Federal courts are watching too: the Judicial Conference's Advisory Committee on Civil Rules has had a draft AI disclosure provision for Rule 11 sitting in committee since March, and today's California vote almost certainly accelerates its adoption.

The Bottom Line

Rule 2.150 is not a guideline — it is a standing order with sanctions attached, effective immediately. Every litigator practicing in California has a homework assignment due before their next filing: choose a compliant AI stack, train the team, update the templates, and read every case before you cite it. The era of quietly using AI to draft court papers is over. The era of openly disclosing it — and being held accountable for what it produces — started today.