In an extraordinary ruling that underscores the growing crisis of generative AI in the legal profession, a federal judge in Mississippi has stricken all four lawyers from a civil case after discovering that both sides had independently submitted court filings riddled with fictitious case citations produced by artificial intelligence. The order, issued on June 8, 2026, by Senior U.S. District Judge Sharion Aycock of the Northern District of Mississippi, terminates the litigation unless the parties can secure new counsel within 60 days and refers every attorney involved to their respective state bar associations for potential disciplinary action. What makes the case particularly alarming is that the opposing counsel were not aware that they were both guilty of the same professional failure, each relying on AI-generated legal research that fabricated precedent from thin air.
The Judge Who Said “Here We Go Again”
Judge Aycock opened her sanctions order with a telling phrase: “This Court is once again forced to address the submission of citations generated by AI hallucination.” The word “once again” carries significant weight. In December 2025, she had already sanctioned lawyers in two separate cases before her court for submitting AI-hallucinated citations. But the present case, Withers v. City of Aberdeen, represents a unprecedented situation: both the plaintiff’s and the defendant’s legal teams had simultaneously filed papers containing fake cases. The court itself discovered the problem when it could not verify any of the cited authorities. Neither side had noticed anything was wrong.
A Fee Dispute That Swallowed Its Own Lawyers
The underlying dispute was, ironically, about legal fees. Plaintiff Tom Withers, a Louisiana-based attorney acting as a pro se litigant, sued the City of Aberdeen, Mississippi, for unpaid legal fees. Withers himself was not sanctioned. The trouble centered on the four lawyers who represented the two sides.
Two attorneys represented Withers: Kathleen Wilson, who drafted the plaintiff’s filings, and Shauncey Hunter Ridgeway, a local Mississippi counsel who signed onto the papers. Two attorneys represented the City of Aberdeen: Kathryn Williams, who drafted the defense filings, and Mark McClinton, another local Mississippi attorney who signed them. All four were sanctioned.
What did they file? Williams’s submissions contained four fabricated cases. Wilson’s contained two more. None of the six cases existed. The court issued an order to show cause on December 10, 2025, demanding explanations from all four lawyers.
How the Timeline Unfolded
Between October and November 2025, both sides filed their competing papers. By December 10, the court had identified the phantom citations and ordered explanations. On January 20, 2026, all four attorneys appeared for a hearing and offered their justifications. The case then took an even more disturbing turn: after the hearing, Wilson filed papers in a separate bankruptcy court in March 2026 that also contained AI-hallucinated cases, leading to additional sanctions in April. On June 8, Judge Aycock issued her final order striking all four lawyers from the Withers case.
Four Lawyers, Four Explanations — None Sufficient
At the January hearing, each attorney offered a different account of how the fabricated cases ended up in their filings. None of the explanations satisfied the court.
Kathleen Wilson, a solo practitioner in Louisiana, admitted that she had used an AI document-generation tool called “First Drafts” to prepare the plaintiff’s filings. She told the court she was unaware that AI could produce fake case citations. “I did not know the term ‘hallucination,'” she claimed. Judge Aycock rejected this as “insufficient and not credible,” noting that by early 2026, ignorance of AI hallucination is no longer a viable defense. The Fifth Circuit had already ruled in Fletcher v. Experian (February 2026) that attorneys cannot plead ignorance of AI risks. Worse, after the January hearing, Wilson continued filing AI-hallucinated cases in a Louisiana bankruptcy court, leading to sanctions there in April. “It is clear that her apology to this Court was not sincere,” Judge Aycock wrote.
Kathryn Williams, a partner at a Texas law firm, admitted using an internal AI legal research tool (the name was not disclosed) to conduct research. The tool was designed for cases in Texas, Louisiana, and Florida, but Williams used it for a Mississippi case, knowing it was not trained on Mississippi law. Her firm had adopted a policy in 2025 requiring verification of all AI outputs. Williams, as a partner, ignored it. Judge Aycock called this “particularly egregious.”
Shauncey Hunter Ridgeway and Mark McClinton, the two local Mississippi counsel, presented a different problem. Neither used AI. Both admitted they had simply signed onto filings prepared by their co-counsel without reviewing them. Ridgeway, representing the plaintiff, and McClinton, representing the city, each failed to exercise the most basic professional responsibility: reading what they signed before submitting it to the court.
What Sanctions Were Imposed
Judge Aycock tailored the penalties to each attorney’s degree of culpability.
Kathleen Wilson: Pro hac vice admission revoked; barred from practicing in the Northern District of Mississippi for two years; $2,500 fine; required to complete continuing legal education on AI ethics.
Kathryn Williams: Pro hac vice admission revoked; barred from the district for two years; $3,500 fine — the higher amount reflecting that she violated her own firm’s AI policy, lied during the hearing about the tool’s geographic scope, and showed a greater degree of bad faith.
Shauncey Hunter Ridgeway and Mark McClinton: Disqualified from the case only; $1,000 fine each. Their failure was negligence, not intentional misconduct.
All four were referred to the bar associations of their respective states for further investigation and potential disciplinary action.
The practical consequence is stark: Withers v. City of Aberdeen now has no lawyers. If new counsel cannot be found within 60 days, the case will be dismissed.
What Is Driving the Surge in AI-Hallucinated Filings
The Withers case is not an isolated incident. It is part of a rapidly escalating pattern that shows no sign of slowing.
Attorney Rob Freund, who tracks AI-hallucination cases, was the first to flag the Withers matter. The broader data comes from Damien Charlotin, a legal data analyst who maintains a worldwide database of judicial decisions involving AI hallucinations. As of mid-2026, the database contains more than 1,500 cases, the vast majority from U.S. courts. On March 31, 2026, alone, 17 court orders were issued in a single day.
Financial penalties are also climbing. In April 2026, an Oregon court imposed an $110,000 fine — the largest in U.S. history for AI-related court filings. At the federal appellate level, the Fifth, Sixth, Seventh, and Ninth Circuits have all issued sanctions for AI-hallucinated citations. In April 2026, Sullivan & Cromwell, one of the most prestigious law firms in the United States, was forced to apologize to a court after submitting AI-fabricated cases.
The warning signs have been visible since the landmark Mata v. Avianca case in the Southern District of New York in 2023, which first brought widespread attention to the problem. Since then, the frequency has accelerated, not diminished.
The Deeper Problem: Convenience Over Verification
The Withers case reveals a professional culture problem that runs deeper than any single tool or vendor. Wilson used “First Drafts,” a specialized legal drafting platform. Williams used an internal firm tool built specifically for legal research. Both were marketed as time-saving solutions for busy lawyers. Both produced convincingly formatted citations that were entirely fabricated.
The technology has improved to the point where its outputs look more polished and authoritative than ever. And that is precisely the danger. As Judge Aycock noted, citing the Mattox case from the Eastern District of Oklahoma, generative AI can produce fluent language, but it cannot produce truth, accuracy, or professional accountability. Those remain the exclusive responsibility of the attorney who signs the filing.
The Fifth Circuit’s Fletcher decision added another critical warning: if an AI tool returns a citation that looks too good to be true, it almost certainly is. Lawyers who treat AI outputs as reliable without verification are not saving time — they are incurring professional risk that can end careers.
The Withers case should serve as a definitive signal to the legal profession. When opposing counsel both file fake cases without realizing it, the problem is no longer about individual bad actors. It is about a systemic failure in how legal work is being produced. The court’s answer in Mississippi was to stop the trial, disqualify every lawyer in the room, and send the matter to the bar. That outcome could become the new normal if the profession does not relearn the oldest lesson in legal ethics: read what you sign before you file it.
