Sullivan & Cromwell Flags BigLaw Risk: AI Hallucinations Spark Reputational Fallout

Sullivan & Cromwell publicly acknowledged AI-generated errors in a court filing due to skipped cite-checking.

Sullivan & Cromwell—one of the most prominent BigLaw firms—recently filed an emergency motion disclosing that AI hallucinations led to citation errors in their Prince Global Holdings Chapter 15 submission. The incident underscores the heightened reputational risks elite firms face as they use AI for drafting legal materials.

• Since mid-2023, more than 120 cases of AI-generated fabricated citations have emerged, with at least 58 in 2025 alone. Judicial sanctions are on the rise, including a $2,500 penalty for attorney Heather Hersh over 21 fake citations in a 5th Circuit brief earlier this year.
• Elite law firms are not immune. In May 2026, Mississippi-based Butler Snow came under court scrutiny after submitting filings with AI-generated fabrications.

The risk extends beyond procedural missteps. A Stanford HAI study found legal AI tools generate falsehoods in roughly 1 out of 6 test queries. This prevalence calls for robust verification protocols—even at the world's most sophisticated firms.

• Judiciary warnings are growing sharper. As one tribunal observed, citing fictitious cases "sends that judge on a fool's errand... not in the interests of justice."
• Despite good intentions, courts warn lawyers: "An attorney who acts with an empty head and a pure heart is nonetheless responsible for the consequences."

The credibility of legal institutions rests on accuracy and integrity. As AI becomes endemic in legal workflows, skipping the human diligence step risks not just sanctions, but public trust—and the standing of even the highest echelons of the profession. Loss of public confidence is a growing concern for both courts and clients.