Sullivan & Cromwell apologized for filing about three dozen AI-hallucinated citations
Sullivan & Cromwell submitted a motion in the bankruptcy case In re Prince Global Holdings Limited containing fabricated case citations and inaccurate passages generated by artificial intelligence. Partner Andrew Dietderich filed an apology letter on April 18, 2026, listing approximately three dozen errors across a three-page attachment, including both AI hallucinations and clerical mistakes. The firm acknowledged it failed to follow internal AI review protocols and stated it was evaluating enhancements to its training and review processes.
An elite law firm let an AI fabricate legal citations and then filed them in federal bankruptcy court without verification.
Key facts
- What
- Sullivan & Cromwell submitted a motion in the bankruptcy case In re Prince Global Holdings Limited containing fabricated case citations and inaccurate passages generated by artificial intelligence.
- Incident date
- Apr 18, 2026
- Who
- Sullivan & Cromwell
- Failure mode
- Hallucination
- AI surface
- Copilot
- Severity
- Medium
What happened
Sullivan & Cromwell filed a motion in the bankruptcy case In re Prince Global Holdings Limited (SDNY Bkr. No. 26-10769) before Chief Judge Martin Glenn that contained AI-generated hallucinations including fabricated case names, case numbers, inaccurate article titles, and imagined passages from real cases. After opposing counsel flagged the errors, partner Andrew Dietderich filed an emergency apology letter on April 18, 2026, listing approximately three dozen errors in a three-page single-spaced attachment. The firm acknowledged it failed to follow internal AI review protocols and stated it was evaluating whether to enhance its training and review processes.
What broke inside the model
- 01 · TriggerA user asks for a fact, a citation, or a figure.
- 02 · Model stepThe model writes a fluent, confident answer.
- 03 · Control gapNothing ties the claim back to a real source.
- 04 · FailureA fabricated fact ships as if it were verified.
- 05 · ConsequenceThe false claim reaches a customer, a court, or the public.
Confidence holds, and even spikes, as the claim detaches from any source.
The AI model generated fabricated legal case citations and imagined passages from real cases, a known hallucination behavior where language models produce plausible-sounding but nonexistent legal references with high confidence. The firm's internal review and verification processes failed to catch these fabrications before the document was filed with the federal court, indicating that human-in-the-loop safeguards were insufficient to detect AI-generated legal fictions.
What it cost
Sources
- Court FilingSullivan & Cromwell Apology Letter for AI Hallucinations (In re Prince USA)websitedc.s3.amazonaws.com
- PressSullivan & Cromwell law firm apologizes for AI 'hallucinations' in court filingreuters.com
- PressA.I. 'Hallucinations' Created Errors in Court Filing, Top Law Firm Saysnytimes.com
Cite this entry
https://failureindex.ai/failures/sullivan-cromwell-apologized-filing-threeAI Failure Index. "Sullivan & Cromwell apologized for filing about three dozen AI-hallucinated citations" (FI-0119). Realm Labs. https://failureindex.ai/failures/sullivan-cromwell-apologized-filing-three (indexed Jun 4, 2026).Data fields CC-BY 4.0, prose citation permitted. Incident ID FI-0119. Full dataset at /data.
Note from Realm Labs, the Index steward
How Realm would have caught this
- Prism
- OmniGuard
- AI Detection & Response (AIDR)
A runtime layer that watches the model's internal state can flag the moment a model commits to a claim it has no support for, and hold or reroute the response before it reaches a user. Realm reads those signals in real time rather than grading the transcript after the fact.