Major Wall Street Law Firm Admits AI Blunders in Court Filing—40 Fabricated Citations Go Undetected

Sullivan & Cromwell, one of America's most prestigious law firms, has apologized to a federal judge after submitting a court filing riddled with approximately 40 incorrect citations and errors stemming from AI hallucinations. This incident reveals a critical vulnerability in how elite legal institutions are deploying artificial intelligence—and it's a wake-up call for the broader professional services sector.

Andrew Dietderich, co-head of Sullivan & Cromwell's global restructuring team, sent a letter Friday to Chief Judge Martin Glenn of the US Bankruptcy Court for the Southern District of New York acknowledging the failure. "We deeply regret that this has occurred," Dietderich wrote, taking personal responsibility for the breakdown in the firm's review procedures. The problematic motion was filed nine days prior to the public admission.

The Scale of AI Hallucinations in Legal Work

This isn't an isolated incident. Damien Charlotin, a legal technologist tracking AI errors, maintains a database recording 1,334 documented cases of AI hallucinations in court filings worldwide—with over 900 occurring in the US alone. Most involve fabricated legal citations, though AI-generated arguments have occasionally created problems too. The pattern suggests law firms are racing to adopt crypto and blockchain-related analysis tools without sufficient guardrails.

Sullivan & Cromwell ranks 30th globally by revenue among law firms and notably represented FTX during its high-profile bankruptcy proceedings—another crypto trading platform that faced regulatory and legal turbulence.

Process Failures Despite Written Policy

Here's where it gets interesting from a crypto market intelligence perspective: Dietderich explicitly stated that Sullivan & Cromwell has formal AI policies in place. These include mandatory citation reviews. Yet the procedures simply weren't followed. "Regrettably, this review process did not identify the inaccurate citations generated by AI, nor did it identify other errors that appear to have resulted in whole or in part from manual error," he acknowledged.

This admission reveals a critical gap between institutional policy and execution—a dynamic traders and investors should understand when evaluating the reliability of AI-driven market analysis tools and crypto trading platforms.

Damage Control and Internal Reckoning

The firm has launched a full internal investigation into how the errors slipped through. They're now "evaluating whether further enhancements to its internal training and review processes are warranted," according to Dietderich. Notably, a competing law firm—Boies Schiller Flexner LLP—actually caught these errors and flagged them, forcing Sullivan & Cromwell's hand.

The fact that rival lawyers had to catch fabricated citations underscores that human oversight remains irreplaceable, particularly in high-stakes litigation where credibility and accuracy define success.