Major US Law Firm Issues Formal Apology for AI-Driven Mistakes in Court Filing
In a significant development highlighting the risks of artificial intelligence in legal practice, Sullivan & Cromwell, one of America's most prominent law firms, has formally apologized to a federal judge for errors in a bankruptcy case motion caused by AI tools. According to a report from the Financial Times, the firm sent a letter to a bankruptcy judge in the US Bankruptcy Court for the Southern District of New York, expressing regret for inaccuracies that included fabricated legal citations and case names generated by artificial intelligence.
Details of the Apology and AI Hallucinations
The apology letter, dated April 18, was written by Andrew Dietderich, founder and co-head of Sullivan & Cromwell's restructuring group. In it, he acknowledged that the firm had become aware of errors in an emergency motion filed in the bankruptcy proceedings of Prince Global Holdings. Dietderich specifically referred to these mistakes as "artificial intelligence hallucinations" and admitted that the firm had failed to follow its established protocols during the document preparation process.
"We sincerely regret the errors in the Motion and the burden they have imposed on the Court and the parties, and I apologize on behalf of our entire team," Dietderich stated in the letter. He emphasized that Sullivan & Cromwell maintains rigorous standards for using AI tools, instructing lawyers to "trust nothing and verify everything." The firm clarified that failing to verify AI-generated output constitutes a direct violation of its internal policies.
Key AI Errors Identified in the Filing
The motion contained several critical inaccuracies attributed to AI, including:
- Fictitious Case Names: The document referenced legal cases that do not exist in reality.
- Fabricated Quotes: It included direct quotes that were never actually spoken or written by any parties involved.
- Errors in Summary: A draft filing erroneously summarized conclusions from other cases, leading to misinformation.
- Non-existent Statutes: The AI incorrectly analyzed or entirely invented provisions within the U.S. Bankruptcy Code.
Background on the AI Tool and Bankruptcy Case
While it remains unclear which specific AI tool was used to generate the faulty motion, Sullivan & Cromwell reportedly holds an enterprise license for OpenAI's ChatGPT. This incident underscores the potential pitfalls of relying on AI in high-stakes legal environments, especially given that the firm's partners typically charge over $2,000 per hour for bankruptcy cases like this one.
The bankruptcy case in question involves Sullivan & Cromwell representing liquidators who are overseeing actions against Prince Group, a Cambodia-based conglomerate. The liquidators were appointed by legal authorities in the British Virgin Islands and are pursuing actions against Prince Group and its owner, Chen Zhi. In 2025, US federal prosecutors charged Zhi with wire fraud and money laundering, accusing him of directing Prince Group's operation of forced-labor scam compounds across Cambodia, which allegedly stole billions of dollars from victims in the United States and worldwide.
This apology highlights growing concerns about the integration of AI in professional sectors, particularly in law, where accuracy and verification are paramount. As firms increasingly adopt AI technologies, this case serves as a cautionary tale about the need for stringent oversight and adherence to verification protocols to prevent similar errors in the future.
