Study of 100 AI-related Court Cases Shows How Lawyer Candor Affects Sanctions

Using Damien Charlotin's publicly available AI Hallucination Cases Database, a team led by Adam Feldman reviewed the 100 most recent U.S. cases where AI-driven errors entered the record. The sample required a clear AI connection, a final judicial outcome, and sufficient detail to assess post-discovery conduct. Findings reveal that nearly every case involving a material error resulted in a serious sanction, but the severity was amplified when attorneys provided deceptive explanations, blamed others, or allowed further inaccuracies after notice.

Prompt admission and comprehensive remediation, such as withdrawing the flawed filing and notifying the court, were linked to markedly lower penalties. Courts favored concrete, voluntary corrective steps, including mandatory source verification, supervisory sign-off, AI-use policies, and training. The analysis underscores that candor cannot undo client harm, yet it can prevent additional professional-responsibility violations.