In a recent court filing in Northern California, a lawyer for Anthropic acknowledged that a citation generated by the company’s Claude AI chatbot was incorrect. The filing indicated that the citation presented “an inaccurate title and inaccurate authors.” Anthropic stated that a comprehensive verification process failed to identify this and other inaccuracies attributed to Claude’s “hallucination” capabilities.
Anthropic’s representative described the incident as an “honest citation mistake” rather than a deliberate fabrication. This admission comes amid accusations from lawyers representing Universal Music Group and other music publishers, who claimed that Anthropic’s expert witness, Olivia Chen, relied on fabricated sources during her testimony. These concerns prompted Federal Judge Susan van Keulen to require a formal response from Anthropic.
The lawsuit underscores an ongoing trend of disputes between copyright holders and tech companies regarding the alleged improper utilisation of creative works to develop generative AI tools. This case is just one of several that have highlighted the challenges of integrating AI into legal practices, illustrated by a recent instance where a California judge rebuked law firms for submitting erroneous AI-generated research. Likewise, an Australian lawyer faced repercussions for relying on ChatGPT for court document preparation, which resulted in faulty citations.
Despite such missteps, the demand for AI in the legal sector continues to grow. Startups like Harvey, which leverage generative AI to streamline legal processes, are reportedly in discussions to secure over $250 million in funding, aiming for a valuation of $5 billion.
This situation illustrates the complexities and potential pitfalls associated with using AI in legal proceedings, raising concerns about its reliability and accuracy while also highlighting its significant appeal in modern legal workflows.
Fanpage: TechArena.au
Watch more about AI – Artificial Intelligence
