Chaney v. Transdev: California Court Fines Plaintiff’s Counsel $2,500 for ‘Bad Faith’ AI-Hallucinated Citations — and Orders Disclosure in Every Active Case | Advocate Prakhar

Case at a Glance

Full Citation	Chaney v. Transdev Services Inc., No. 2:24-cv-10761-ODW (C.D. Cal. Apr. 28, 2026)
Court	United States District Court, Central District of California (Judge Otis D. Wright, II)
Date	April 28, 2026
Category	AI Hallucination — Bad Faith Sanctions; Concealment Finding
Jurisdiction	United States — California (Central District)
AI Tool Used	Generative AI (admitted after initial denial)
Outcome/Sanction	$2,500 monetary sanctions ($500 + $1,000 + $1,000 per instance); disclosure declaration required in every active Central District case; bar referral threat if non-compliant

Background

Tekoma Chaney worked as a bus driver for Transdev Services Inc. from March 2019 until January 2020, when she was terminated. She later brought a civil rights and labour law action in the Central District of California, asserting fifteen causes of action against Transdev and her supervisor Jamie Ogilvie. Her case was complicated by a 2023 bankruptcy in which she failed to disclose her pending employment claims — a judicial estoppel issue that ultimately barred most of her claims. The court granted both defendants’ motions for summary judgment. Separately, the court also addressed the AI misconduct of Chaney’s attorney, referred to as “Gabriel” in the order.

The AI Issue

Gabriel cited a case called “In re Shubert” in Chaney’s opposition briefs — once in his response to Ogilvie’s motion for summary judgment, and twice in his response to Transdev’s motion. “In re Shubert” does not exist. When first confronted, Gabriel described it as a “copy-and-paste error.” Only when the court issued a formal order to show cause did Gabriel admit he had used generative AI without verification. The court found two independent bases for bad faith: first, Gabriel knew about the risks of AI hallucination and used it anyway without verifying the citations; second, he made “multiple and repeated attempts to obfuscate and minimize” his AI misuse — initially concealing it, then failing to acknowledge that the same fabricated citation appeared in a second brief he did not initially disclose.

What the Court Decided

• Summary judgment was granted for both Transdev and Ogilvie — Chaney’s undisclosed bankruptcy claims barred her from asserting them [judicial estoppel].
• Gabriel’s use of AI-generated hallucinated citations three times (across two briefs) constituted bad faith under the court’s inherent authority [bad faith finding, not mere negligence].
• Attempting to frame AI hallucination as a “copy-paste error” before finally admitting AI use is itself sanctionable conduct — concealment aggravates the underlying violation [concealment as aggravator].
• $500 for the first use of In re Shubert; $1,000 for each of two uses in the Transdev brief — total $2,500 [per-instance graduated sanctions].
• Gabriel must file a disclosure declaration in every case where he is an appearing attorney in the Central District — explaining his AI use in this case and the court’s bad faith findings [system-wide disclosure obligation].
• For judges in the Central District before whom Gabriel does not currently appear, he must email the declaration to their chambers — proactive disclosure rather than waiting for future appearances [prophylactic requirement].
• Failure to comply will result in contempt proceedings and potential referral to the California State Bar [escalation threat].

“Any attorney diligently fulfilling his or her ‘duty to keep abreast of the changes in the law and its practice, including the benefits and risks associated with relevant technology’ should be aware of the grave peril to which they subject themselves by relying on generative-AI without verification.”

— Judge Otis D. Wright, II, Chaney v. Transdev, April 28, 2026

The India Angle

Indian Law Equivalent

The “duty to keep abreast of changes in the law and its practice, including relevant technology” that Judge Wright cites from the California Rules of Professional Conduct has a direct analogue in the Bar Council of India Rules: advocates are expected to maintain competence in their field of practice. Section 49(1)(c) of the Advocates Act, 1961 empowers the Bar Council to make rules for professional conduct — and several State Bar Councils have begun examining whether AI competence forms part of an advocate’s professional duty.

Bar Council Rules

Rule 22 (no false statements), Rule 15 (fair conduct), and Rule 14 (dignity of court) are all engaged. But the Chaney case adds a new dimension: Rule 22’s prohibition on false statements explicitly extends to the conduct of explaining those statements to the court. When Gabriel first described AI hallucination as a “copy-paste error,” he was making a false representation to the court about the cause of his misconduct — a second-order violation that converted a negligence issue into a dishonesty finding.

Practical Advice for Indian Advocates

• If AI misconduct is identified in your filing, disclose it immediately and fully — attempting to minimize or recharacterize it will be treated as aggravating conduct and will significantly increase any sanction.
• The California court’s per-instance graduated sanction ($500 for the first use, $1,000 for each repeat) is a model that Indian courts may adopt — the financial risk of submitting the same hallucinated citation multiple times is multiplicative, not linear.
• The system-wide disclosure obligation imposed on Gabriel — requiring him to inform every judge in the district about his AI misconduct — illustrates how AI-related penalties can affect an attorney’s entire practice, not just the case where the misconduct occurred.

Quick Takeaways

• Trying to explain away AI hallucination as a “copy-paste error” turns negligence into bad faith.
• Per-instance graduated sanctions mean the same fake citation in two briefs costs twice as much.
• A system-wide disclosure obligation is a court-imposed permanent reputational consequence that affects all active cases.

---

Deep Dive: The Concealment Problem — Why Cover-Ups Make AI Hallucination Cases Much Worse

The Chaney case is the clearest illustration in 2026 of what happens when an attorney attempts to manage — rather than disclose — an AI hallucination problem after it is discovered. Gabriel’s progression from “copy-paste error” to “I used AI without verifying” tracks a pattern that Judge Wright dissects methodically: the initial declaration buried the AI admission among eleven paragraphs; only two of those paragraphs addressed the core issue. Gabriel mentioned only one brief, not both. Only after the court’s show-cause order forced full disclosure did the complete picture emerge.

Judge Wright’s reasoning on bad faith is worth careful reading. He identifies two independent bases. First, Gabriel knew about AI hallucination risks — courts have been widely reporting on this issue since at least the 2023 Mata v. Avianca decision — and used AI anyway without verification. This is the ordinary negligence-to-willfulness upgrade that courts increasingly apply as AI hallucination becomes common knowledge. But the second basis is more significant: Gabriel’s attempts to conceal and minimize. These are affirmative acts of misrepresentation to the court — not about the underlying case facts, but about Gabriel’s own professional conduct. When an attorney misrepresents to a court why they made an error, they are engaging in exactly the kind of dishonesty that professional conduct rules are designed to prevent.

The disclosure obligation imposed on Gabriel is innovative in its architecture. It does not wait for future violations. It requires Gabriel to proactively inform every judge in the Central District of California — in their chambers email — of what he did and what the court found. This transforms the typical “you’re on notice” warning into an active dissemination of information about Gabriel’s professional conduct. Every judge in the district will have, in their chambers file, a document from Gabriel explaining his AI misconduct. Future judges will not need to research whether Gabriel has a history of these issues — they will already know. For Indian advocates, this case raises the question of whether Indian courts can and should adopt similar mechanisms: requiring advocates who have submitted hallucinated citations to proactively inform other courts in which they are appearing, rather than relying on those courts to independently discover the prior misconduct.