Pro se litigants are using artificial intelligence tools to prepare briefs and pleadings. The problem is that they are using artificial intelligence chatbots which make up citations to spurious cases. AI programs don’t have emotions. They cannot feel shame. They have no fear of embarrassment. Like humans they love to tell the user exactly what he or she wants to hear. Another district court has admonished a pro se litigant in the case Butler v. Fidelity Management, No. 25-cv-9337 (S.D.N.Y)
The court politely admonishes the plaintiff:
Troublingly, Plaintiff letter contains five citations to cases that clearly do not support the proposition for which they are cited; quotations from three cases that do not appear in the cases cited; and one citation to a nonexistent judicial opinion.[1] I believe that these issues are due to Plaintiff’s use of generative AI tools. See, e.g., Romero v. Goldman Sachs Bank USA, No. 25-CV-2857 (GHW), 2025 WL 1916119, at *1-2 (S.D.N.Y. June 25, 2025) (cautioning that “citations, quotations, and holdings that have been generated by generative artificial intelligence may be hallucinations”). But no matter what the cause, the “presentation of false citations, quotations, and holdings by a party . . . to the Court is sanctionable conduct.” Id. at *2; see also Park v. Kim, 91 F.4th 610, 615 (2d Cir. 2024) (per curiam) (observing that the “attempt to persuade a court or oppose an adversary” by relying on “non-existent precedent generated by ChatGPT” is an “abuse of the adversary system”). Had this filing been made by a lawyer, I would consider imposing sanctions. Because Plaintiff is not a lawyer, I decline to pursue the matter at this time. Nevertheless, Plaintiff is cautioned that presentation of false citations, quotations, or holdings in the future may lead me to impose sanctions. See, e.g., Advani v. App. Term, 2nd Jud. Dep’t, No. 25-CV-1627 (JMF), 2025 WL 2201065, at *3 (S.D.N.Y. Aug. 1, 2025); Zeng v. Chell, No. 19-CV-3218 (JGK), 2024 WL 3360570, at *6 (S.D.N.Y. July 9, 2024) (“Were it not for the plaintiff’s pro se status and the proper judgment dismissing this case, a further inquiry” into the plaintiff’s potential use of AI “would be appropriate.”); Reilly v. Conn. Interlocal Risk Mgmt. Agency, No. 25-CV-640 (VDO), 2025 WL 1726366, at *3 (D. Conn. June 20, 2025) (“Whether the issues with [the pro se plaintiff’s] filings are the result of artificial intelligence or some other mistake, the Court cautions [the plaintiff] to ensure that future submissions to any court contain only accurate representations.”).
While I understand the difficulties of litigating pro se and the reasons why a pro se litigant might choose to use AI-generated filings, Plaintiff is admonished that pro se litigants are obligated to review the cases cited in their filings to confirm that they exist, stand for the proposition for which they are cited, and contain any quotations attributed to them. See Bhagat v. Shah, No. 24-CV-1424 (VEC) (RFT), 2026 WL 925605, at *2 (S.D.N.Y. Apr. 6, 2026).
Comment: I’m not sure how we can address this issue humanely while allowing pro se litigants access to the courts.