Sullivan & Cromwell Cites AI Hallucinations in Court Filing Error
Sullivan & Cromwell LLP apologized to a New York bankruptcy judge on April 18 after discovering that a court submission contained fabricated and misquoted case citations. The firm attributed the errors to artificial intelligence hallucinations in its legal research process.
The mistake appeared in a filing related to In re: Prince Global Holdings Ltd., a bankruptcy case. The firm did not specify which AI tool generated the false citations or how extensively the system was used in preparing the document.
What This Means for Law Firms
The incident underscores a basic constraint: AI systems generate plausible-sounding but false information with confidence. They do not actually verify citations against case databases. They predict text based on patterns in training data.
For attorneys, this creates a verification problem that cannot be solved by using AI more efficiently. Every citation, every case reference, every factual claim that AI generates requires independent confirmation against authoritative sources.
Sullivan & Cromwell's status as a top-tier firm demonstrates that size and resources do not prevent these errors. The mistake happened despite the firm's access to legal research platforms and experienced lawyers.
The Broader Pattern
This case is one of several instances where AI systems have produced false legal citations in court filings. Each incident reveals the same underlying issue: AI cannot be trusted to perform legal judgment on its own, even when tasked with narrow, seemingly straightforward work.
Attorneys remain responsible for every factual claim in their filings. That responsibility cannot be delegated to a system that hallucinates.
Learn more about AI for Legal professionals and how to integrate these tools safely into legal workflows, or explore the AI Learning Path for Paralegals to understand where human oversight is essential.
Your membership also unlocks:
