Anthropic Attributes Legal Citation Error to Claude AI’s Formatting Mistake
Anthropic recently addressed concerns over an inaccurate citation included in a legal filing related to its ongoing dispute with major music publishers. The filing, submitted by Anthropic data scientist Olivia Chen on April 30th, was intended to defend the company against allegations that copyrighted lyrics were improperly used to train its AI chatbot, Claude.
During a hearing, an attorney representing Universal Music Group, ABKCO, and Concord described one of the sources in Chen’s filing as a “complete fabrication,” suggesting it was hallucinated by the AI. Anthropic’s defense attorney, Ivana Dukanovic, responded by confirming that the source was genuine and acknowledged that Claude was used to format the legal citations.
What Went Wrong?
While a manual check corrected some errors, Anthropic admits that certain wording mistakes in the citation went unnoticed. The company clarified that although the publication title, year, and source link were accurate, the citation contained an incorrect title and authors’ names. Dukanovic emphasized the error was not an intentional fabrication of authority.
Anthropic apologized for the confusion caused, calling the citation error “an embarrassing and unintentional mistake.”
AI Citation Errors in Legal Contexts Are Increasing
This incident highlights a growing problem with AI-generated citations in legal documents. Just last week, a California judge criticized two law firms for submitting a supplemental brief filled with fabricated materials generated by AI, which were never disclosed to the court.
Additionally, a misinformation expert admitted that ChatGPT hallucinated citations in a legal filing submitted in December, underscoring the risks of relying on AI tools without thorough human verification.
Practical Takeaways for Legal Professionals
- AI can assist with formatting citations but should not replace manual verification to ensure accuracy.
- Legal teams must maintain transparency about the use of AI tools in filings to avoid credibility issues.
- Understanding AI’s limitations with citation generation is essential to prevent unintentional errors that may impact legal proceedings.
For legal professionals interested in responsible AI use and improving their skills in this area, training resources are available. Comprehensive courses on AI tool integration and prompt engineering can offer practical guidance on minimizing risks in legal documentation. Explore relevant AI training courses for legal professionals to stay informed and prepared.
Your membership also unlocks:
