Legal AI Hallucination Leads to Wrongful Judgment: New York Court Awards $1.2 Million in Damages

The U.S. District Court for the Southern District of New York issued a landmark ruling on September 4: because the AI legal research tool LexiSearch used by defense counsel generated fabricated case citations, an $8 million contract dispute resulted in an erroneous judgment. The court ordered the AI tool's developer to pay $1.2 million in damages to the affected party.

The case unfolded as follows: defense counsel used LexiSearch for case law research during trial preparation. The tool cited six precedents supporting the defense's position, but upon verification by plaintiff's counsel, four of those precedents simply did not exist—they were LexiSearch "hallucination" outputs. After discovering the issue, the judge overturned the original judgment and initiated contempt of court proceedings against the AI tool's developer.

LexiSearch's developer argued in court that its terms of service clearly stated "all outputs require human verification" and therefore should not bear full responsibility. But the judge rejected this defense, ruling that "when a professional tool's error rate reaches 67%, disclaimers cannot serve as a basis for immunity."

The American Bar Association issued a statement on September 7 recommending that all members independently verify every citation when using AI legal tools.