DC Appeals Court: Blames Deutsche Bank AI Hallucinations, Strikes Brief

• The DC Court of Appeals struck down a Deutsche Bank subsidiary's appellate brief due to AI-generated fake citations.
• Lawyer Loishirl Hall admitted Google's generative AI produced four nonexistent case citations in the brief.
• The court held all attorneys who signed the brief responsible, not just the primary author.
• The matter has been referred to the District of Columbia Bar's Office of Disciplinary Counsel.
• The court emphasized that while AI use is not discouraged, attorneys must use it correctly and responsibly to avoid misleading the judiciary.

The court unequivocally stated that a brief containing fabricated citations, while appearing to advocate, ultimately misleads the judiciary, consumes valuable time and resources, and creates confusion, leaving the client without credible representation.

The District of Columbia Court of Appeals has issued a stark warning regarding the use of generative artificial intelligence in legal practice, striking down an appellate brief filed by a Deutsche Bank subsidiary. The unanimous three-judge panel found that the brief, submitted in an ongoing mortgage foreclosure lawsuit, contained multiple citations to nonexistent cases, which were later attributed to AI hallucinations. This decision, reported on September 4, 2026, underscores the critical need for attorneys to rigorously verify all AI-generated content.

The court's action came after Loishirl Hall, a lawyer who joined DC Affordable Law Firm as a fellow in August 2022, and was formerly with McCabe, Weisberg & Conway, who represented the Deutsche Bank subsidiary, admitted that four of the problematic citations had been fabricated by Google's generative AI tool. Hall acknowledged that these fake citations should never have appeared in a court filing. The ruling serves as a significant precedent, highlighting the potential for legal tech malpractice AI when not properly supervised.

While Loishirl Hall took responsibility for the AI-generated errors, the District of Columbia Court of Appeals extended culpability beyond a single individual. The panel's per curiam opinion explicitly stated that every firm attorney who signed the flawed brief bears some measure of responsibility for its contents. This finding emphasizes a collective attorney responsibility generative AI output, signaling that oversight cannot be delegated solely to the user who initially inputs the query.

Despite the severe consequences in this instance, the court clarified that its intention is not to deter or prohibit lawyers from utilizing artificial intelligence. On the contrary, the opinion noted that attorneys who choose to eschew at least an understanding, if not direct use, of AI do so at their own peril. However, the court strongly underscored the paramount importance of employing AI tools correctly and responsibly, ensuring that technological assistance does not compromise the integrity of legal submissions.

The court unequivocally stated that a brief containing fabricated citations, while appearing to advocate, ultimately misleads the judiciary, consumes valuable time and resources, and creates confusion, leaving the client without credible representation. The panel concluded that the damage is done once the court is forced to investigate the veracity of citations, diverting attention from the substantive legal arguments. Consequently, the Deutsche Bank subsidiary's brief was struck in its entirety.

In a further step, the matter has been formally referred to the District of Columbia Bar's Office of Disciplinary Counsel for review, indicating potential sanctions for the attorneys involved. In a separate concurring opinion, Senior Judge Stephen Glickman of the District of Columbia Court of Appeals agreed with the outcome but suggested that the per curiam opinion did not sufficiently delve into the broader issues surrounding AI use, attorney responsibilities, and the full spectrum of potential sanctions. This case serves as a critical warning for lawyers and compliance officers regarding the ethical and professional responsibilities when using generative AI for legal research.

This case is a critical warning for lawyers and compliance officers regarding the ethical and professional responsibilities when using generative AI for legal research. It underscores the necessity of rigorous verification of all AI-generated content, especially case citations, to avoid disciplinary action and ensure professional credibility. Firms must establish clear policies and training on responsible AI tool usage to mitigate the risk of 'hallucinations' leading to sanctions.

Source: Original reporting via Reuters