The federal court in Tacoma recently excluded an expert whose report contained AI-generated “hallucinated” citations to academic articles.
The court’s order in LeDouxx v. Outliers, Inc., 2026 WL 2408808 (W.D. Wash. Aug. 18, 2026) (unpublished), was the latest of several opinions in the same case addressing inaccurate citations in both legal papers filed by the plaintiff’s lawyer and in associated expert reports. The month before, the court sanctioned the lawyer for “‘dozens of inaccurate factual and legal citations across at least five different filings.’” 2026 WL 2137370 at *1 (W.D. Wash. July 24, 2026) (unpublished). In that order, the court observed (at *4) that some of the expert reports contained similar problems but deferred specific consideration of those at that time. Still earlier, the court had excluded a different set of experts on other procedural grounds but also noted that at least one of their reports included “hallucinated” citations. 2026 WL 76377 at *11 n.6 (W.D. Wash. Jan. 9, 2026) (unpublished).
The lawyer eventually conceded that she had a role in preparing the expert reports and that the inaccurate citations were generated by two AI tools that she claimed she did not understand how to use (and that the experts apparently did not verify themselves). 2026 WL 2137370 at *4.
The most recent order came in the context of a defense summary judgment motion. The lawyer, who represented the plaintiff, offered an expert report in opposition to the defense motion. In excluding the expert’s report, the court found that the “hallucinated” citations rendered the expert’s opinions inherently unreliable under Federal Rule of Evidence 702 governing expert testimony. The court concluded that the expert’s “multiple hallucinated citations ‘shatter … his credibility with this Court.’” 2026 WL 2408808 at *7 (citation omitted). Having excluded the expert, the court found that the plaintiff had not come forward with sufficient evidence to withstand the defense motion and entered summary judgment dismissing the plaintiff’s claims with prejudice.
So far, most decisions involving AI-generated “hallucinated” citations have come in the context of legal filings and resulting sanctions against the lawyers involved. The latest chapter in LeDoux, however, points to a more fundamental risk when it comes to experts: if an expert is excluded, the party may not be able to prove their case. That, in turn, puts a premium on ensuring that citations included in expert reports are accurate and suggests that conversations with experts about their use of AI tools in preparing their reports should now be prudent practice.


