A federal district court judge in Mississippi has disqualified four attorneys after discovering fabricated case citations in legal briefs submitted by both sides in a lawsuit.
The case, *Withers v. Aberdeen*, is pending in the Northern District of Mississippi before Judge Sharion Aycock, who issued a detailed 23-page opinion on the matter.
The dispute stems from a contractual pay disagreement between Tom Withers III and the city of Aberdeen, drawing in two sets of attorneys who each independently relied on unverified AI-generated research.
Judge Aycock applied Rule 11, which governs attorney conduct and grants sanction authority, alongside the court’s inherent powers and local rules in determining appropriate penalties.
Plaintiff’s attorney Kathleen Wilson claimed she was unaware that AI could produce hallucinated cases, an explanation Aycock described in her order as “insufficient and incredulous.”
Wilson’s position was further weakened when she admitted she had filed other pleadings containing undiscovered fictitious citations, including in courts outside Mississippi, even after the hallucinations in this case had already come to light.
The court also noted that Wilson allowed her local counsel, who had not prepared any of the pleadings, to take the lead in addressing the fictitious citations rather than taking responsibility herself.
Wilson had her pro hac vice admission revoked, was barred from appearing in the district for two years, fined $2,500, and ordered to complete continuing legal education.
Texas-based defence attorney Kathryn Y. Williams, admitted pro hac vice to represent the municipality, received a matching two-year ban and a $3,500 fine for “blindly relying” on an AI research tool.
Williams’ firm had an AI policy requiring all citations to be verified, and the court found it particularly egregious that as a “partner and leader in her firm,” she disregarded that policy entirely.
Williams also attempted to argue that while the cited cases did not exist, the legal principles they represented were valid, a no harm, no foul defence that Judge Aycock similarly rejected.
She further damaged her credibility by misrepresenting her schedule in what appeared to be an attempt to avoid the hearing Aycock had demanded on the hallucination issues.
The local counsel fared little better: Shauncey Hunter Ridgeway admitted she had not read the brief before signing it, while Mark McClinton said he read the brief but did not verify any of the citations.
Neither local attorney knew the out-of-state lawyers had used AI, and both stated they had not used AI tools themselves in preparing their contributions to the case.
Ridgeway and McClinton, who had each sponsored the temporary admissions of Wilson and Williams respectively, were fined $1,000 each and removed from the case entirely.
Aycock referred all four attorneys to appropriate bar disciplinary authorities, extending the consequences of the hallucination failures well beyond the immediate proceedings.
“Unfortunately for the attorneys, I think the judge gets it exactly right,” said Ben Cooper, a professor at the University of Mississippi School of Law who studies AI and serves on the Mississippi Bar Association’s Ethics Committee.
The ruling arrives as courts across the United States continue to grapple with the growing problem of attorneys submitting AI-generated content without adequate verification of the sources cited.
U.S. District Judge Henry T. Wingate of the Southern District of Mississippi previously admitted that his own staff used AI to draft a flawed court order, illustrating the breadth of the problem across the legal system.