Skip to content
Submit incident
AII-1000-71931
Well documentedThree or more independent publishers report the core facts.Supporting sources are pending verification.

Sullivan & Cromwell apologizes for AI hallucinations in bankruptcy court filing

Sullivan & Cromwell, a major Wall Street law firm, apologized to a federal bankruptcy judge for a court filing containing AI-generated errors including fabricated case citations and misquoted authorities. The errors were discovered by opposing counsel and the firm acknowledged its AI safeguard policies were not followed.

18 April 2026 Β· Sullivan & Cromwell Β· Reliability and hallucination
18 April 2026
AII-1000-71931
Quick facts
Category
Reliability and hallucination
Country
United States
Date
18 April 2026
Drafted by
Raidu Incident Detection

What happened

Sullivan & Cromwell submitted a motion in US Bankruptcy Court in New York containing errors generated by artificial intelligence tools. Andrew Dietderich, a partner at the firm, wrote an apology letter to Judge Martin Glenn on April 18 acknowledging the errors.

The motion contained inaccurate citations, fabricated case citations, misquoted authorities, and non-existent legal sources. The errors were discovered by lawyers from the opposing firm Boies Schiller Flexner.

Reported impact

The firm's AI safeguard policies were not followed and the errors made it past the firm's internal citation review process. Dietderich stated the firm takes responsibility for the failure and committed to submitting a corrected filing.

Index analysis

The incident reflects that AI hallucinations producing fabricated legal citations have become more common in legal work since 2023, according to legal researcher Damien Charlotin.

Governance controls

Sullivan & Cromwell stated it has comprehensive policies regarding AI use and safeguards designed to prevent such errors, but acknowledged these policies were not followed in this instance.

Classification

Category
Reliability and hallucination
Industry
Business processes and support services
Country
United States
Harm status
Actual harm
Intent
Unintentional
Timing
Post-deployment
Harm types
Reputational
Harmed
Business
Failure modes
Fabricated citation, Hallucination
Consequences
Public apology
AI systems
AI tool (unnamed)

Sources and evidence

Sources not yet published

This record was researched and written by the Index from original reporting. Its sources are still being verified by an editor, and they are listed here once that check is complete. Until then the record carries no reviewer.

How sources are verifiedSuggest a source for this incident