Sullivan & Cromwell just pulled a bankruptcy court filing. The reason? AI tools generated fake case citations and bogus information.
This isn’t some small firm. Sullivan & Cromwell is one of America’s most prestigious law practices.
Andrew Dietderich heads the firm’s restructuring practice. He acknowledged the errors to Judge Martin Glenn. The firm’s verification protocols exist for exactly this reason. They were bypassed in a case involving Prince Group.
The AI-generated motion contained 28 incorrect citations. Court reports detail non-existent quotes. Completely fabricated.
Defense lawyers from Boies Schiller Flexner caught the errors. They objected. They called the firm’s post-objection corrections prejudicial. They’re requesting a status conference to address the mess.
Dietderich acknowledged what he called “AI hallucinations.” He expressed regret to Judge Glenn at New York Southern Bankruptcy Court.
The incident hits hard. Sullivan & Cromwell handles complex financial restructuring cases. They’ve got stated policies requiring verification of AI-generated outputs. Those policies didn’t prevent this.
The firm pledged stricter AI verification procedures. Too late for this case.
The broader question: how are legal practices integrating AI tools? Clearly without sufficient safeguards.
This adds to growing scrutiny around AI’s reliability in legal settings. Courts are testing AI for administrative tasks. They remain wary of its limitations. The legal community is grappling with efficiency versus accuracy. Judicial integrity can’t be compromised.
The opposing counsel’s objections highlight a key concern. AI-generated errors in official court filings create prejudice. They complicate proceedings. Corrections don’t erase the damage. The defense team’s request for a status conference signals this isn’t a simple clerical error.
For Sullivan & Cromwell, this is a reputational moment. The firm’s known for extensive work in finance and restructuring. The admission stings.
The case underscores a harsh reality. Even top-tier legal practices with established verification policies can fall victim to AI hallucinations. The safeguards have to be followed. Always.
AI adoption is accelerating across the legal industry. This case is a cautionary example. Efficiency gains can’t come at the expense of traditional verification processes.
The outcome of any court review could influence how law firms approach AI integration going forward. The entire industry is watching.
