Kaur v. Desso

Gurpreet Kaur v. Captain Joel Desso, et al., U.S. District Court for the Northern District of New York, July 9, 2025, 9:25-cv-726 (AMN), Doc. 24 (Judge Anne M. Nardacci). Copy of primary document: `~/workspace/goals/ai-sanctions-case-monitoring/files/sources/D-70-kaur-v-desso.pdf`.

Dennis Desmarais used the generative AI tool Claude Sonnet 4 to "assist with research and drafting" a supplemental brief in an emergency habeas matter. The brief contained machine-fabricated legal quotations. Desmarais admitted he knew AI tools are known to "hallucinate" but made no attempt to check the quotations, and even after the government flagged the fabrications he took no remedial action before the show-cause hearing.

Violation of Federal Rule of Civil Procedure 11(b)(2) and (b)(3), found sua sponte under the subjective bad faith standard (conscious avoidance equivalent to knowledge). Sanctions: a $1,000 monetary sanction payable to the Clerk of Court within 14 days, and proof to the court of participation in a CLE-credited program related to the use of artificial intelligence in legal writing by September 1, 2025, and proof by the same date that he had served the Petitioner with a copy of the order. Source URL: the primary PDF (Charlotin documents archive: /documents/581/Kaur_v._Desso_NDNY_9_July_2025.pdf, https://www.damiencharlotin.com/documents/581/Kaur_v._Desso_NDNY_9_July_2025.pdf).

Knowing AI hallucinates means you must verify, not skip verification; knowledge of the risk makes the failure worse. In a time crunch, seek an extension instead of filing unverified AI output. Withdraw and correct a filing the moment an opponent flags fabrications.

• damiencharlotin.com: https://www.damiencharlotin.com/documents/581/Kaur_v._Desso_NDNY_9_July_2025.pdf