Part of the AI Regulation News hub.
A federal judge in California declined a fine, ordered a lawyer to report the court's AI sanctions order to the State Bar herself by 13 November 2026, and told her not to use the non-mandatory reporting form
The counsel in the case asked for a fine small enough to stay under the State Bar's mandatory reporting threshold. The court called that request troubling, and structured the sanction so the report happens anyway.
Bottom line: A binding order in this case only, not precedent for other courts. The court imposed non-monetary sanctions rather than a fine, ordered counsel to self-report the order to the State Bar of California, and told her not to use the non-mandatory reporting form.
Who this affects: California litigators and plaintiff-side employment and education counsel who file in the Central District, and any lawyer weighing a negotiated, below-threshold fine in an AI-misuse show-cause proceeding.
Issue date: Order dated 26 October 2026. Self-report to the State Bar by 13 November 2026. Proof of self-reporting filed with the court by 25 November 2026.
What changed: No new rule. The court used a form of sanction, an order to self-report, in place of a fine. It reasoned that a monetary sanction would not prevent harm to clients, opponents, the court and the public.
Analysis: The sanction is built around the reporting threshold. Counsel told the court she would pay what it set but asked that the amount be under one thousand dollars to avoid mandatory State Bar reporting. The court said it found that request troubling and chose a remedy that does not depend on a dollar figure. Our inference: when a court wants the Bar informed, a smaller fine may not buy the lawyer silence.
Primary sources: Order Imposing Non-Monetary Sanctions, C.D. Cal., Dkt. 26, 26 October 2026 (PDF, Charlotin database copy)
- Instrument (EN)
- Order Imposing Non-Monetary Sanctions and Referral to State Bar of California of Attorney Angelina Mariscal (Dkt. 18), Civil Minutes - General, Dkt. 26
- Case number
- 5:26-cv-03777-MWC-SP
- Authority
- United States District Court, Central District of California; the Honorable Michelle Williams Court
- Jurisdiction
- United States, federal court; California State Bar reporting
- Status
- Order entered in chambers. We did not check the docket for later filings.
- Bindingness
- Binds the attorney named in the order. Not precedent for other courts.
- Sanction
- Non-monetary: self-report to the State Bar of California; not to use the form for non-mandatory reporting; file proof of self-reporting
- Issue date / next deadline
- 26 October 2026; State Bar self-report by 13 November 2026; proof to court by 25 November 2026
- Editorial Note
- Informational analysis for working professionals, not legal advice. Confirm how any rule applies to your situation with qualified counsel.
- Primary source
- https://www.damiencharlotin.com/documents/3382/Chechitelli_v._Etiwanda_School_District_USA_26-10-2026.pdf
What the order requires
Three things, set out in the conclusion. Attorney Angelina Mariscal is ordered to self-report the order to the State Bar of California, for the purpose of determining whether she violated the California Rules of Professional Conduct and the State Bar Act, by no later than 13 November 2026. She is ordered not to use the form for non-mandatory reporting. And she is ordered to file proof of self-reporting with the court by 25 November 2026.
A footnote gives the Bar's reporting form for judicial sanctions. The order does not say the form is the required route; it says only that she is not to use the one for non-mandatory reporting.
How the court got there
The case began as a state-court action against the Etiwanda School District, removed to federal court in March 2026. The district moved to dismiss in August, the plaintiff did not respond, and on 28 September 2026 the court granted the motion and issued an order to show cause on monetary sanctions and a State Bar referral, citing concern about misuse of AI.
The court's account of the complaint is specific. It says most of the six causes of action are not recognized under existing law or provided for in the statutes cited, that the California Education Code does not give a private cause of action for damages, and that some causes of action, such as Coercion and Fear-Mongering, are nonsensical and nonexistent. The September order to show cause, which the court quotes, said these bear the hallmarks of AI hallucinations, a phrase it took from another court.
Counsel's sworn declaration said generative AI was used as a research and drafting-assistance tool for portions of the complaint, and that she verified that the cited statutes existed. The court records that at the 2 October hearing she first denied using AI, was reminded of her declaration, and then acknowledged using AI to draft the initial complaint. According to the order she also said she had "Googled" for some cases and used AI to complete her research.
What the court said about duties
The court grounded its reasoning in the California Rules of Professional Conduct on competence and diligence and in Rule 3.1 on claims not warranted under existing law. It quoted the Ninth Circuit's Malkeet Lnu v. Blanche for the proposition that, at a bare minimum, the duties require lawyers to review generative AI output for fabrications and basic accuracy. We have not opened that decision.
It found that counsel conceded she did not read the statutes on which she based the causes of action, and that she did not represent to the court that she had a good-faith argument for extending or changing the law. It recorded her statement that her intent was to hold the school district's feet to the fire, and that she conceded there was nothing at filing to say there was a claim.
It also rejected her statement that she had withdrawn the complaint, noting that no withdrawal or request for leave to amend was filed, and it noted that the declaration itself bore indicia of unverified AI use, because she appeared unfamiliar with it at the hearing.
Why not a fine
The court said monetary sanctions would not prevent harm to clients, opponents, the court and the public. It treated her request for a below-threshold fine as troubling, and found that her actions in the case raise a substantial question about her honesty, trustworthiness or fitness as a lawyer.
That last sentence paraphrases the court's finding in the order. It is not our characterization of counsel, and the order is a judicial finding made after a hearing, not a State Bar disciplinary result. Nothing in the order says what the Bar will do.
The calendar
The order sets two dates, and they sit close together. The self-report is due 13 November 2026, and proof that she did it is due to the court on 25 November 2026. The order states no consequence for missing either date, and we did not look for one elsewhere.
For anyone tracking the case, those are the only two events the order schedules. Whether either happens is a docket question we did not check.
What we did not verify
We opened and read the full nine-page order. The text contains the case number 5:26-cv-03777-MWC-SP, the date October 26, 2026, the document footer Document 26 Filed 10/26/26, and the heading stating the order covers Dkt. 18. What we say about the complaint, the declaration and the hearing is the court's description of them.
We did not open the complaint, the September 28 order to show cause (Dkt. 18), the Mariscal declaration (Dkt. 20), the hearing record (Dkt. 21), the motion to dismiss, or the Ninth Circuit's Malkeet Lnu decision or the State Bar's practical guidance that the court quotes. The order does not say specific cases were fabricated; it points to non-cognizable and nonexistent causes of action, and we make no claim beyond that.
We did not check whether counsel has self-reported, filed proof, or sought relief from the order, and we did not verify the State Bar's reporting thresholds or forms; counsel's reference to a one-thousand-dollar figure is her request as the court records it. We make no claim about any State Bar proceeding.
In this order, the way a sanction is structured mattered as much as its size. A lawyer in an AI-misuse show-cause proceeding who is trying to stay under a reporting threshold may find the court choosing a different remedy that reaches the Bar anyway. The order's reasoning points to the plain steps: read the authorities you cite, check what AI produces, and make sure a sworn declaration matches what you will say at the hearing.
Source File
Open the order and confirm the case number 5:26-cv-03777-MWC-SP and the footer Document 26 Filed 10/26/26; then find Part IV (Duty to Report Attorneys Who Engage in Professional Misconduct) and Part V (Conclusion) on pages 8 and 9 for the 13 November and 25 November 2026 dates and the instruction not to use the non-mandatory form.
The Court finds monetary sanctions will not prevent harm to clients, opponents, the Court, and the public, caused by Mariscal's irresponsible use of artificial intelligence in her practice. ยท Judge Michelle Williams Court, C.D. Cal., 26 October 2026
FAQ
What did the court order?
That the attorney self-report the order to the State Bar of California by 13 November 2026, not use the form for non-mandatory reporting, and file proof of self-reporting with the court by 25 November 2026.
Was there a fine?
No. The order is titled as imposing non-monetary sanctions. The court said monetary sanctions would not prevent harm to clients, opponents, the court and the public.
Why does the form matter?
According to the order, counsel asked for a monetary sanction below one thousand dollars to avoid mandatory reporting to the State Bar. The court ordered her not to use the non-mandatory reporting form. We did not verify the Bar's reporting rules, so we do not state the threshold ourselves.
Does the order say the lawyer used AI to invent cases?
Not in those terms. The court describes non-cognizable and nonexistent causes of action, records counsel's statements that she used AI to draft the complaint and complete research, and does not, in the text we read, find specific cases fabricated.
Related briefings
Sponsored Training
Practical AI training for regulated professionals, built around verification, documentation and a defensible process. See the courses.