$2900 in Sanctions for AI Hallucinations in Filings by Self-Represented Litigant

From Monday’s decision in Lothamer Tax Resolution, Inc. v. Kimmel, by Chief Judge Hala Jarbou (W.D. Mich.):

This order comes in response to pro se Defendant Paul Kimmel’s pattern of submitting misleading filings to the Court. In an opinion issued August 29, 2025, the Court observed that one of Kimmel’s filings contained the error-filled citations that are a hallmark of content generated by artificial intelligence (AI), and warned Kimmel that he must verify the accuracy of his citations in the future. In a subsequent order issued November 19, 2025, the Court noted that Kimmel’s recent filings still cited dozens of misrepresented or fake cases, and ordered Kimmel to explain why he should not be sanctioned for this conduct. Kimmel responded on November 24, 2025, claiming that the inaccurate citations resulted from editing errors and stylistic mistakes rather than from AI. For the reasons explained below, the Court finds that Kimmel violated Rule 11 and will impose monetary sanctions in the amount of $2,900….

Kimmel acknowledges that “[s]everal citations or quotations in [his] Objection were inaccurate.” He does not admit to using AI, though; he avers “that some of those mistakes arose from relying on secondary summaries or search-engine case pages rather than the official opinion.” Ultimately, he characterizes his errors as “occasional technical mistakes [that] are unavoidable, but … never intentional.” He states that he will take steps in the future to verify his citations and avoid misrepresenting cases. However, as explained below, the Court finds Kimmel’s response lacking.

As an initial matter, Kimmel’s response itself contains misleading citations…. As to the errors in his previous filings, Kimmel claims that he lacks the time to address each one, but he does not generally dispute the Court’s finding that his brief is filled with misleading citations. Instead, he notes generally that the incorrect quotations were largely caused by “quoting from secondary sources … that paraphrase the law rather than reproduce the official court opinions themselves.” However, Kimmel does not point to the particular sources he used for each quotation or provide any other evidence for this assertion. Kimmel also attempts to explain two specific errors identified by the Court, but—as explained below—even that minimal explanation is lacking….

Even if the Court were to accept all of the claims in Kimmel’s response—that he did not understand how to use quotation marks, quoted paraphrases from secondary sources, and accidentally added “ghost” reporter citations—these claims do not explain most of his (over two dozen) errors….

To be clear, the Court does not view formatting mistakes or occasional accidental misquotes as sanctionable conduct. But Kimmel’s briefs are so riddled with incorrect citations that none of the case law in them can be relied upon, and checking them for accuracy wastes the time of opposing counsel and the Court. Although Kimmel denies that his fake citations come from generative AI, “[t]he Court finds this response inadequate and not credible,” given that Kimmel “fails to explain where or how he found or created the fictitious case[s] and quotation[s].” Regardless, “[t]he Court need not make any finding as to whether [Kimmel] actually used generative AI to draft any portion of his [filings]” in order to sanction the use of fake citations. Indeed, if Kimmel has created the fake citations himself, that would suggest an intentional attempt to mislead the Court—a more serious transgression than carelessly filing AI-generated documents. Ultimately, wherever the fake citations originated, “[t]he filing of complaints, papers, or other motions without taking the necessary care in their preparation is a[n] … abuse of the judicial system, subject to … sanction.” …

It is true that Kimmel is proceeding pro se, and lacks the experience or resources available to an attorney. And the Supreme Court has acknowledged that, under Rule 11, “the legal inquiry that can reasonably be expected from a party [before filing a brief] may vary from case to case.” Nevertheless, Rule 11 “imposes on any party who signs a pleading, motion, or other paper … an affirmative duty to conduct a reasonable inquiry into the facts and the law before filing.” And “[p]ro se plaintiffs are not exempt from Rule 11 sanctions simply because they are not represented by counsel.” The frequency of Kimmel’s falsehoods indicates that he failed to make a reasonable inquiry into the veracity of his citations, and this failure is sanctionable regardless of whether he is represented by counsel.

“A sanction imposed under [Rule 11] must be limited to what suffices to deter repetition of the conduct or comparable conduct by others similarly situated.” The Court finds that the proper sanction in this instance is a monetary fine of $100 for each false citation, to be paid to the Court. Cf. Ali v. IT People Corp., Inc. (E.D. Mich. Sept. 19, 2025) (sanctioning pro se party $200 for each false citation). In the list above, the Court identified [29] instances of false citations…. Kimmel’s response to the order to show cause does not take issue with the Court’s list of erroneous citations. Thus, the Court will impose a sanction of $2,900. If Kimmel continues to submit filings containing misleading or fictitious citations, the Court will double the monetary sanction per violation for each additional time that it has to address the issue.

The post $2900 in Sanctions for AI Hallucinations in Filings by Self-Represented Litigant appeared first on Reason.com.