The Expert RecordThe standard for expert witnesses using AI

AI in the Courtroom · Fabricated citations and facts in filings

Cervantes v. Bianco (Supreme Court of California)

In an original writ proceeding over ballots seized by the Riverside County Sheriff, the sheriff's preliminary briefing misquoted and misstated one of the court's own decisions. The firm apologized and withdrew the argument. The Supreme Court of California disapproved the improper use of AI tools and referred the matter to the State Bar.

Primary document readAI use disapproved; matter referred to the State Bar September 24, 2026

What happened

Cervantes v. Bianco, No. S295866, was an original petition for a writ of mandate in the Supreme Court of California. The petitioners challenged the Riverside County Sheriff's seizure of roughly 650,000 ballots from the November 4, 2025 special election. The merits concerned the Elections Code provisions on who may handle and count voted ballots. This entry addresses only one point in the case: what happened in the sheriff's briefing.

In his preliminary briefing, Sheriff Chad Bianco challenged the petitioners' standing. According to footnote 7 of the court's opinion, that argument rested on "misquoted passages from, and an incorrect recitation of our holding in," Common Cause v. Board of Supervisors (1989) 49 Cal.3d 432, a decision of the same court. The law firm responsible apologized and withdrew the standing argument. The opinion does not name the firm.

What the court did

On September 24, 2026, the court granted the petition in part, directing the sheriff to return the ballots to the Registrar of Voters, and denied the rest without prejudice. Because the sheriff no longer contested standing, the court discussed it only briefly. In footnote 7 it addressed the briefing error directly:

"We disapprove of the improper use of artificial intelligence tools, which apparently contributed to the significant error that occurred here and remind counsel of attorneys' professional duties of competence and candor to the court. (See Rules Prof. Conduct, rules 1.1, 3.3.) We also refer the matter to the State Bar of California to determine whether discipline is warranted. (See canon 3D(2) of the California Code of Judicial Ethics.)"

The court imposed no monetary sanction. The referral cites canon 3D(2) of the California Code of Judicial Ethics, the canon addressing a judge's response to a lawyer's professional misconduct.

Where it stands

The referral is to the State Bar of California, which decides whether discipline is warranted. No State Bar outcome was located as of October 8, 2026. The writ proceeding itself is concluded as to the relief granted.

Cervantes v. Bianco joins State v. Sandoval and Payne v. State as matters in which a state supreme court acted on AI errors in briefing before it. It arrived a week before SB 574 was chaptered, which will bar California filings containing citations the responsible attorney has not personally verified. It also sits alongside the Court of Appeal's sanctions in Noland v. Land of the Free and the Supreme Court's own order in Kjoller v. Superior Court. The error here was not a nonexistent case. It was a real decision quoted inaccurately and described wrongly, the harder kind of error to catch because the citation checks out.

The obligation it engaged

The conduct engages Clause 3 of the Standard, independent authority, in the form most often missed. The Standard requires that every quotation and characterization be confirmed against the source, not merely that the source exist. A real case with an invented passage passes a citation check and fails a reading.

For an expert the lesson is the same. A system that summarizes a regulation, a standard or a published study can attribute words to a real document that the document does not contain. Opening the source and finding the passage is the only check that catches it.

Sources

What could not be confirmed

The law firm responsible for the briefing is not named in the opinion and was not identified. The opinion says AI tools 'apparently contributed' to the error; it does not identify a tool. The copy read was the AI Hallucination Cases database's hosted copy of the court's PDF.

About this entry

This entry was checked against the primary document listed in the sources. The entry for Cervantes v. Bianco records conduct and what the tribunal did. It does not characterize a holding and it is not legal advice. Last checked October 8, 2026. More matters like this one are collected under fabricated citations in filings. If something here is out of date, write to [email protected].

Top