An AI Screw-Up By... Sullivan & Cromwell?

Key Takeaways

• S&C filed a bankruptcy motion with inaccurate AI‑generated citations.
• Firm apologizes to U.S. Bankruptcy Court chief judge for the errors.
• Errors were flagged by rival Boies Schiller Flexner, a known AI‑fail champion.
• S&C also advises OpenAI on safe AI deployment, highlighting conflict.
• Over 1,300 AI hallucination cases tracked across courts, indicating industry‑wide issue.

Pulse Analysis

The latest AI mishap at Sullivan & Cromwell underscores how quickly generative tools have infiltrated high‑stakes legal work. In a Chapter 15 case, the firm relied on an AI‑drafted brief that cited nonexistent authorities, prompting an emergency apology to the court. While the error was caught by Boies Schiller Flexner—a firm already notorious for its own AI blunders—the episode adds another data point to the growing "AI Hall of Shame" that tracks faulty filings across the nation’s courts. For a firm that markets itself as a trusted advisor to OpenAI on ethical AI deployment, the slip is especially conspicuous.

Law firms are now wrestling with the tension between efficiency gains and the risk of AI hallucinations. Tools such as RealityCheck by BriefCatch are emerging to automatically flag dubious citations before they reach the docket, but adoption remains uneven. The incident at S&C illustrates that reliance on AI without robust human oversight can erode client confidence and expose firms to sanctions. As litigation budgets tighten, firms must balance cost‑saving incentives against the potential for costly retractions, reputational damage, and even malpractice claims.

Looking ahead, the legal sector is likely to see tighter standards and possibly formal guidelines governing AI use in filings. Firms that already counsel technology companies on responsible AI—like S&C—may need to internalize those principles more rigorously. Investment in layered verification workflows, staff training, and specialized AI‑audit software will become a competitive differentiator. Ultimately, the episode serves as a cautionary tale: the promise of AI in law is real, but without disciplined checks, even the most prestigious firms can stumble.