Law firm Gordon Rees escapes sanctions over AI errors in bankruptcy case

Reckless use of generative AI in legal proceedings has resulted in the personal reprimand of attorney Cassie Preston of the Gordon Rees Scully Mansukhani law firm by a U.S. bankruptcy judge. 

In a Chapter 11 bankruptcy case filed by Jackson Hospital & Clinic in Alabama, U.S. Bankruptcy Judge, Christopher Hawkins, decided not to impose formal sanctions on Gordon Rees Scully Mansukhani, even after the firm submitted court filings containing AI-generated legal citations that turned out to be made up.

Courts are treating fabricated AI-generated citations as serious ethical and procedural violations worthy of sanction.

Judge reprimands lawyer over careless AI use

In his ruling, Judge Hawkins said that Gordon Rees avoided reprimand because the firm had “taken reasonable steps” to address the risks associated with using generative AI for legal work. The firm had already reimbursed the opposing parties with more than $55,000 in attorneys’ fees relating to the faulty submissions.

To prevent future problems, the firm also changed its internal rules to include a compulsory “cite-checking” process whenever AI tools are used to generate research or draft documents.

While the firm as a whole was spared, Judge Hawkins officially reprimanded Gordon Rees’ former attorney Cassie Preston, who was responsible for the filings on behalf of creditor Progressive Perfusion.

Preston explained that she had some difficult personal circumstances at the time of the submission. Despite that, the judge found that her conduct “constituted an abuse of the bankruptcy process.”

Hawkins’ detailed opinion filled 32 pages and included the fact that Preston “doubled down, tripled down, and quadrupled down on arguments unsupported by the authorities cited.”

As part of her punishment, Preston is required to distribute the judgment and the court’s opinion to all her clients, opposing counsel, and judges in any ongoing cases.

Growing concern over AI in legal filings

Courts across the U.S. are increasingly flagging fake legal citations and misstated legal authority generated by AI tools. In a Puerto Rico FIFA-law case for example, two lawyers were fined over $24,400 for submitting filings with more than 50 faulty or non-existent citations.

Three attorneys from the firm Butler Snow were publicly reprimanded by a federal judge in Birmingham, Alabama, Anna Manasco, for using ChatGPT to generate legal citations that turned out to be entirely made up.

The judge described their conduct as “recklessness in the extreme.” The lawyers were disqualified from the case and, just like Preston, were ordered to share the court’s order and opinion with all their clients, opposing counsel, and judges in their other cases.

According to reports by The Washington Post, courts in the U.S. have documented at least 95 incidents of lawyers submitting filings with fake or made-up cases generated by AI. Such errors erode trust in the legal system.

Steve Puiszis, general counsel at the firm Hinshaw, says that lawyers must always independently verify any case law, even if it was generated by AI.