Sanctions for Undisclosed AI Use Underscore Risks of Generative Tools in Litigation

According to an EDRM article by Michael D. Berman, in Lacey v. State Farm General Ins., the court imposed sanctions following the undisclosed AI use that led attorneys to submit a brief containing inaccurate, AI-generated legal citations twice. Despite being alerted to errors in the original filing, counsel resubmitted a revised version that still included multiple bogus citations. The Special Master found nine of 27 citations in the ten-page brief were flawed, including at least two that referenced non-existent cases and others that misquoted legal authority.

Berman says the misconduct stemmed from one attorney’s use of AI to generate an outline, which was incorporated into the final brief without proper verification or disclosure. Unaware of the AI-generated content, attorneys at a second firm failed to cite-check the submission before filing. Even after the Special Master flagged citation anomalies, the revised filing still included AI-generated errors, without disclosing the technology’s role in the drafting process.

The Special Master concluded that the collective conduct amounted to reckless advocacy with an improper purpose and imposed sanctions under Federal Rules of Civil Procedure 11 and 37, as well as the court’s inherent authority. While individual attorneys were spared personal sanctions due to their candor and remorse, the plaintiff was ordered to pay $26,100 in Special Master fees and $5,000 in defense costs. Several briefs were stricken, and related discovery relief was denied.

Lacey delivers a clear warning: generative AI tools may be tempting shortcuts, but undisclosed AI use, especially without basic validation, risks sanctions, reputational damage, and adverse rulings. Courts expect lawyers to maintain rigorous standards, regardless of the tools employed. Cite-checking remains a non-delegable duty. AI is not an excuse for professional negligence.