Chaney v. Transdev Services Inc.
U.S. District Court, Central District of California · C.D. Cal. · California bar guidance
Conduct
Counsel, who admitted drafting parts of a brief with LexisNexis+ Protégé, cited a fake case three times, then disclosed one use in an improper filing as a copy-and-paste error.
Consequence
$2,500 sanction for bad faith, plus a declaration of the AI misuse and bad-faith findings filed in every C.D. Cal. case where he appears and emailed to every other judge.
Lesson
Here the court held that failing to verify AI output is itself bad faith, and sanctioned a partial, buried disclosure as a separate act of bad faith.
Verified September 29, 2026
- Citation
- Chaney v. Transdev Services Inc., No. 2:24-cv-10761-ODW (AJRx), Order Granting Motions for Summary Judgment [31] [32] and Imposing Sanctions (C.D. Cal. Apr. 28, 2026) (Wright, J.), ECF No. 46
- Decided
- April 28, 2026
Summary
Tekoma Chaney sued her former employer, Transdev Services Inc., and Jamie Ogilvie for employment-related claims. In opposing Ogilvie's summary-judgment motion, her counsel, Gavril T. Gabriel, cited "In re Shubert, 799 F.3d 1124, 1129 (9th Cir. 2015)" for the proposition that property abandoned by a bankruptcy trustee reverts to the debtor. After Ogilvie's reply pointed out that the case was fake, and after the motions were under submission, Gabriel filed a declaration that, among paragraphs the court treated as an improper sur-response, called the Shubert cite an "inadvertently made citation" meant for a different real case. Judge Otis D. Wright, II found that In re Shubert "is not a real case"; its reporter cite leads to an unrelated D.C. Circuit decision. In response to an order to show cause, Gabriel admitted using LexisNexis+ and its Protégé drafting feature to draft portions of the brief. While preparing its ruling, the court found that the same fabricated case also appeared twice in Chaney's opposition to Transdev's motion, which Gabriel had not disclosed in either declaration.
- AI tool:
- LexisNexis+ and its drafting feature Protégé (counsel admitted using them to draft portions of the brief)
- Sanction amount:
- $2,500 payable to the Court ($500 for one use of the fabricated case and $1,000 for each of two more); declaration of the AI use and the bad-faith findings to be filed in every C.D. Cal. case where counsel appears and emailed to every other C.D. Cal. judge's chambers
What sanction did the court impose?
Relying on its inherent authority to sanction bad faith, the court made two bad-faith findings. First, given how well documented sanctions for AI hallucinations now are, it held that "the failure to verify citations after using generative-AI rises to the level of bad faith." Second, it found that Gabriel's "multiple and repeated attempts to obfuscate and minimize his generative-AI misuse also warrants sanctions for bad faith conduct," including burying the admission in an improper filing, presenting the case as a copy-and-paste error, and omitting its two other appearances. The court sanctioned him $2,500, payable by May 28, 2026, and ordered him to file by May 12, 2026, in every Central District case in which he appears, a declaration explaining how he used generative AI in this case and describing the court's bad-faith findings, and to email it to the chambers of every other Central District judge. The court warned that failure to comply will lead to further disciplinary proceedings, including contempt and referral to the State Bar. It separately granted summary judgment to both defendants for lack of standing.
Why does Chaney v. Transdev Services Inc. matter for law firms using AI?
Judge Wright’s sanctions section begins, “Unfortunately, the Court must, once again, discuss sanctions related to the use of generative-artificial intelligence.” The fabricated case, In re Shubert, surfaced first in Ogilvie’s reply. Counsel’s first response came in a post-submission declaration that the court found attempted “to improperly rehabilitate his use of certain case law” and to supplement his opposition, with the Shubert admission placed in the middle. Only after an order to show cause did counsel admit that he used LexisNexis+ and its Protégé drafting feature.
The court’s first bad-faith finding is notable: it reasoned that sanctions for AI hallucinations are now so well documented that an attorney who keeps up with technology, as California Rule of Professional Conduct 1.1 requires, “should be aware of the grave peril” of relying on generative AI without verification, so that the failure to verify “rises to the level of bad faith.” The second finding rests on concealment: the court discovered on its own that the fabricated case appeared twice more in the opposition to Transdev’s motion, which counsel mentioned in neither declaration.
The court also explained its choice of remedy. It found monetary sanctions necessary but said they “have proved inadequate deterrence generally,” and so paired a per-use fine with district-wide disclosure of its findings. Firms documenting compliance may wish to consider that, in this case, the court calculated the fine citation by citation and required the lawyer to carry the bad-faith findings into every other case he has in the district.
Implications for your firm
Operational steps a firm reading this case may wish to consider documenting. Strategic and rule-application calls belong to your firm's attorneys.
- Verify output from AI drafting features built into established research platforms. Counsel admitted using LexisNexis+ and its Protégé drafting feature to draft portions of the brief that cited the fabricated case.
- Document every appearance of a problem citation before disclosing it. The court faulted counsel for disclosing one use of the fake case and not the two in a second brief.
- Review how a correction is presented. The court treated a disclosure buried in an unauthorized sur-response, framed as a copy-and-paste error, as an apparent attempt to hide the misconduct.
- Consider the reach of notice sanctions; here counsel must disclose the findings to every judge in the district.