A California federal judge recently struck a portion of an expert report submitted by Anthropic PBC in a copyright lawsuit due to errors attributed to an AI-generated miscitation. Magistrate Judge Susan van Keulen took this action after discovering that the report included a reference to an article hallucinated by the AI, a mistake which was not caught during the manual citation check reportedly performed by Anthropic’s counsel, Latham & Watkins LLP. This incident highlights ongoing challenges in incorporating AI technologies into legal practices, particularly around the reliability of AI-generated content.
The ruling, issued in the U.S. District Court for the Northern District of California, involved a case brought by music publishers accusing Anthropic of copyright violations. In her May 23 order, Judge van Keulen questioned how the error went unnoticed, noting that it affected the credibility of the report. She ordered the removal of the paragraph citing the non-existent article.
The incident underscores the importance of rigorous checks in the use of AI in the legal field, where the credibility of information is paramount. While AI has the potential to streamline certain processes, such cases illustrate the need for human oversight to ensure accuracy and reliability. More details about this ruling can be found on Bloomberg Law.