<img src="//bat.bing.com/action/0?ti=5189112&amp;Ver=2" height="0" width="0" style="display:none; visibility: hidden;">

    The Lawletter Blog

    CIVIL PROCEDURE: Sanctions for Attorney who Repeatedly Submitted Error-Riddled AI-Generated Briefs

    Posted by Lee P. Dunham on Fri, Feb 27, 2026 @ 09:02 AM

    The Lawletter Vol. 51 No. 1

    Lee Dunham, Senior Attorney

          On February 5, 2026, Judge Failla of the United States District Court for the Southern District of New York issued an Opinion and Order in Flycatcher Corp. Ltd. v. Affable Avenue LLC, 24 Civ. 9429 (KPF), 2026 U.S. Dist. LEXIS 23980 (S.D.N.Y. Feb. 5, 2026), imposing severe Rule 11 sanctions on attorney Steven A. Feldman for repeatedly filing apparently AI-generated submissions with false citations, including an initial motion to dismiss and several more briefs in response to the court’s Order to Show Cause why the initial motion should not be dismissed.

          The Opinion is notable for its detailed (and often bitingly funny) description of the kinds of outlandish bad-faith shenanigans that will invite the rare remedy of terminal sanctions. It also serves as a warning that, if courts ever considered the novelty of Large Language Models (“LLMs”) to be an excuse or mitigating factor for attorneys who misuse the technology, that kind of leniency can no longer be expected.

          The court described Feldman’s initial brief in support of the motion to dismiss as “peppered with false citations,” including “at least thirteen cases that did not exist, and eight cases that did exist but did not contain the quotes . . . attributed to them.” The brief also contained a three-page “introduction” as well as a largely duplicative “preliminary statement,” which made opposing counsel suspicious that Feldman had used an LLM such as ChatGPT to generate the brief. When opposing counsel sent him an email notifying him of the errors and demanding that he withdraw the brief, Feldman offered a “convoluted excuse” and claimed that he would replace the erroneous citations, but offered no timeline to do so. Opposing counsel notified the court by letter when Feldman failed to withdraw or correct the brief, and the court ordered Mr. Feldman to show cause why the brief should not be stricken from the docket and Rule 11 sanctions imposed on him.

          Feldman appears to have relied on an LLM again to draft a response to the show cause order, which the court described as being “noteworthy for its conspicuously florid prose,” including “an extended quote from Ray Bradbury’s Fahrenheit 451,” “metaphors comparing legal advocacy to gardening,” and a reference to the “sacred trust” signified by the styluses carried by the scribes of the “ancient libraries of Ashurbanipal.” “Needless to say,” these rhetorical flourishes “raised the Court’s eyebrows.” Astonishingly, the response also contained a misquotation, which appeared to be AI-generated, from Mata v. Avianca, Inc., 678 F. Supp. 3d 443 (S.D.N.Y. 2023), a seminal case about AI-generated “hallucinated” citations. The court scheduled a conference for Mr. Feldman to explain himself, to “give him the opportunity to — as he puts it — ‘prove [himself] worthy to carry the stylus once more in service of justice and truth.’” The court emphasized that all lawyers “must know how to verify that a case exists on Westlaw without the added benefit of AI tools,” and that Feldman was “not excused from this professional obligation by dint of using emerging technology.”

          Thereafter, Feldman requested leave to submit a proposed reply brief. Opposing counsel notified the court of yet more citations to nonexistent and apparently “AI hallucinated” cases in the proposed brief.

          At the conference, Feldman was evasive and offered convoluted explanations that failed to satisfactorily explain the problematic submissions. The transcript can be found at https://storage.courtlistener.com/recap/gov.uscourts.nysd.633195/gov.uscourts.nysd.633195.223.0.pdf. The court concluded that he “misused AI in three separate filings, which resulted in the submission of nonexistent cases and misattributed quotes.”

           The court cited Park v. Kim, 91 F.4th 610, 615 (2d Cir. 2024), for the proposition that Rule 11 requires, at a minimum, that attorneys must “read, and thereby confirm the existence and validity of, the legal authorities on which they rely.” The court also noted that it had inherent authority to impose sanctions for bad faith conduct. Because Feldman “violated Rule 11 repeatedly and brazenly, despite multiple warnings from the Court and fellow counsel,” the court elected to impose terminal sanctions by entering a default judgment as to his client, and ordered opposing counsel to submit an application for attorneys’ fees.

    Topics: civil procedure, artificial intelligence, attorney ethics

    New Call-to-action
    Free Hour of Legal Research  for New Clients

    Subscribe to the Lawletter

    Seven ways outsourcing your legal research can empower your practice

    Latest Posts