A civil lawsuit over unpaid fees has collapsed in the federal court in the Northern District of the US state of Mississippi. Attorney Tom Withers represented the city of Aberdeen in the contract dispute. Withers, however, did not represent himself.
As the US online magazine 404 Media reports, the problem lay in the inadequate work of the lawyers of both parties to the dispute. Both Withers’ and Aberdeen’s lawyers had used generative artificial intelligence to support their legal arguments.
In doing so, they included the output of the software in their written documents without checking. The result: Both sides presented the court with supposed precedents that did not exist in reality to support their respective positions. The AI had simply invented the alleged court rulings.
Judge Sharion Aycock then immediately broke off the hearing and withdrew the mandate for this case from all four lawyers involved. In addition, the court imposed fines of between $1,000 and $3,500 and, in some cases, two-year bans from appearing in the court district. Plaintiff Withers himself was spared the sanctions because he was not held responsible for his legal counsel’s errors.
AI systems argue against each other
The case highlights the risk of so-called hallucinations, in which language models generate plausible-sounding but entirely fictitious facts. According to court documents, both parties effectively pitted two chatbots against each other without subjecting the generated output to human review.
“In an era of rampant, untested AI use in the legal industry, this case is a prime example of the risk inherent in waving results through unseen,” Judge Aycock criticized the conduct. This systematic failure to monitor content justifies the penalties imposed.
Lack of control mechanism for text generators
This is not the first time that language models have caused problems in legal disputes. Last year, a case at the federal court in the US state of New York caused a stir because the plaintiff relied on ChatGPT for research.
However, the current development in Mississippi shows a new level of escalation, as basic legal due diligence was violated on both sides. Tools based on artificial intelligence accelerate processes, but do not replace the final technical inspection by human experts.
Editorial recommendations
${content}
${custom_anzeige-badge}
${custom_tr-badge}
${section}
${title}
The lawyer Rob Freund, who first discovered the matter and shared it on the short message platform X, described the process as a “comedy of AI errors”. He pointed out that the clients had essentially paid for a language model to argue against itself.
Although the sanctions imposed do not mean a complete withdrawal of approval, they do have a massive impact on the professional reputation of those involved. The case clearly shows that the unchecked outsourcing of responsibility to algorithms represents an enormous risk.
Top Article
${content}
${custom_anzeige-badge}
${custom_tr-badge}
${section}
${title}
