Skip to content

Incident Tracker / United States (federal)

LeDoux v. Outliers, Inc., No. 3:24-cv-05808-TMC (W.D. Wash.)

·US District Court for the Western District of Washington, Tacoma·US-FEDSanctions

In a Western District of Washington product-liability suit, Judge Tiffany Cartwright fined plaintiff's counsel USD 3,000 for dozens of ChatGPT- and Claude-generated false citations, calling the conduct tantamount to bad faith. On 18 August 2026 the court excluded three expert reports that relied on the same AI-fabricated academic citations and granted summary judgment dismissing the case.

Court
US District Court for the Western District of Washington, Tacoma
Date
2026-07-24
Who used the AI
Lawyer
AI tool
ChatGPT and Claude (per counsel's sworn declaration)
Conduct
Plaintiff's counsel Jocelyn Stewart submitted dozens of inaccurate legal and factual citations across at least five filings, including a nonexistent case (Berg v. Chevrolet Motor Div.) and a fabricated version of RCW 7.72.010(2). She also gave two experts an AI-generated citation table containing nonexistent academic articles, and a third expert's report contained its own hallucinations.
Outcome
Sanctions
Penalty
US$3,000
Regulatory outcome
No separate regulatory disposition recorded
Status
Verified against listed sources
Last checked
2026-09-03

What was fabricated or misused

  • RCW 7.72.010(2) — proposed amended complaint quoted nonexistent text and changed the definition of “manufacturer”
  • “supplying a component part” — phrase alleged as manufacturing activity, but the term does not exist in RCW 7.72 and RCW 7.72.010(2) does not include someone who supplies a component part
  • Over a dozen quotes from expert reports, including James Kababick’s “post-production degradation of stimulant compounds,” were not in the cited transcripts or reports
  • United States v. Hankey, 203 F.3d 1160 (9th Cir. 2000) — quoted phrase “unsupported speculation masquerading as scientific fact” was not in the decision
  • Berg v. Chevrolet Motor Div., 84 Wn.2d 102, 108, 524 P.2d 226 (1974) — nonexistent case; correct citation was Baughn v. Honda Motor Co., 107 Wn.2d 127, 142, 727 P.2d 655 (1986)
  • Fair v. King Cnty., 2025 U.S. Dist. LEXIS 79249 — quoted phrases “jurisdictional bar” and “the better rationale for striking...” were not in the decision
  • In re Washington Mut., Inc. Sec., Derivative & ERISA Litig., 694 F. Supp. 2d 1192, 1223 (W.D. Wash. 2009) — quoted “unclean hands” language was not in the decision
  • In re Washington Mut., 2011 WL 1158387, at *2 (W.D. Wash. Mar. 25, 2011) — quoted “less stringent” language was not in the order
  • Mr. Kababick’s and Dr. Shippee’s expert reports cited ten nonexistent or misattributed academic articles; one example misnamed a 2019 Kranenburg article as a 2022 Forensic Chemistry article by Bynum
  • HSS Enters. v. AMCO Ins., 2008 U.S. Dist. LEXIS 31659 (W.D. Wash. Apr. 16, 2008) — quoted “harsh sanction” language was not in the decision

What the court said

“In November and December 2025, Plaintiff Joann LeDoux submitted multiple filings containing dozens of inaccurate factual and legal citations which appeared “hallucinated” by a generative artificial intelligence tool.”
US District Court for the Western District of Washington, Tacoma, ECF 265, p. 1
“Ms. Stewart concedes that she violated Rule 11 by failing to properly review her submissions to this Court, resulting in dozens of false citations.”
US District Court for the Western District of Washington, Tacoma, ECF 265, p. 5
“Here, Ms. Stewart’s actions warrant significant sanctions. Ms. Stewart submitted dozens of false citations throughout four briefs and two expert reports.”
US District Court for the Western District of Washington, Tacoma, ECF 265, p. 10
“The Court finds that the hallucinated citations undermine Dr. Holguin’s entire report and that the report is not reliable under Daubert.”
US District Court for the Western District of Washington, Tacoma, ECF 269, p. 12

Timeline

  1. Defendants raised false-citation concerns at partial summary judgment argument
  2. Court ordered Stewart to show cause under Rule 11
  3. Stewart filed a sworn declaration naming Claude and ChatGPT
  4. Court imposed a $3,000 personal sanction and certification duties
  5. Deadline for Stewart to certify service of the sanctions order on LeDoux
  6. Court excluded expert reports and granted summary judgment

Why this case matters

LeDoux adds an expert-evidence branch to the usual fake-authority pattern. Mata and Lnu focused on briefs; here, AI-generated article tables moved into expert reports, then undercut Daubert reliability and summary judgment. The sanction order also shows why later cooperation does not erase a Rule 11 problem: the court credited Stewart’s admissions and changed practices but still found bad faith because the volume of obvious errors consumed court and party resources. It is a useful comparator for Mitchell, where a claim chart rather than case law contained false intrinsic-record citations.

Practice note

Counsel should not give experts AI-generated citation tables unless each article has been opened and matched to the report’s propositions. For any brief, keep a citation table showing the source, quoted passage and check date. If an opponent flags errors, withdraw or amend all affected filings, explain how each error arose, and stop filing new AI-assisted citations until the review is complete.

Primary sources for this incident

Related regulation records for this incident

No related regulation record has been linked to this incident yet.

Similar incidents on the record

Cite this record

SafeLegalAI Global Legal AI Incident Tracker, “LeDoux v. Outliers, Inc., No. 3:24-cv-05808-TMC (W.D. Wash.)”, safelegalai.com/tracker/ledoux-v-outliers (accessed 2026-10-07). Data: CC BY 4.0.

Report an issue with this record →