Wall Street Elite’s AI Blunder: Fabricated Citations Expose Cracks in Legal Tech Reliance

On April 9, 2026, Sullivan & Cromwell, a prominent law firm advising major clients like OpenAI and Donald Trump, filed an emergency motion in the U.S. Bankruptcy Court in Manhattan. The motion, regarding the collapse of the Prince Group, was riddled with errors, including bogus citations, misquoted statutes, and nonexistent cases. Opposing counsel flagged approximately 40 inaccuracies, prompting the firm to issue a correction and apologize to Judge Martin Glenn. The errors were attributed to AI "hallucinations," despite the firm holding an enterprise license for OpenAI's ChatGPT and having policies in place for its use. This incident mirrors previous high-profile cases, such as Mata v. Avianca in 2023, where lawyers were fined for citing fake cases, and Morgan & Morgan's troubles in Wadsworth v. Walmart in 2025. The legal community is grappling with the dual pressures of AI's promise of efficiency and the risks of unverified output. Courts are increasingly enforcing Rule 11 compliance, and while some firms are implementing stricter training and verification tools, the frequency of AI-related errors continues to rise. As one analysis noted, "A fake opinion is not 'existing law,'" and the reliance on AI without proper human oversight remains a significant liability for legal professionals.

(Source：Webpronews)