Federal Court Sanctions Attorney for AI‑Generated Fake Case


Hi TCPAWorld!

Courts across the country have seen a surge in orders issuing sanctions for fictitious, “hallucinated” legal citations resulting from reliance on generative AI. Now, generative AI sanctions have entered the TCPAWorld!

In Matt McCormick v. Texakoma Financial, Inc., 4:25-cv-773 (E.D. Tex. June 11, 2026), U.S. District Judge Amos L. Mazzant issued financial sanctions and publicly reprimanded an attorney for violating Federal Rule of Civil Procedure 11 for, amongst other improper uses of generative AI, citing a case that did not exist.

Defendant Texakoma Financial Inc. moved for summary judgment after Plaintiff Matt McCormick alleged to receive three phone calls in violation of the TCPA. In response, Plaintiff filed a Motion for Summary Judgement Response (“Response”) in which Defendant followed up with a Reply. In its Reply, Defendant noted, amongst other issues, that Plaintiff’s Response contained a fictitious case, fictitious quotes, and unsupported propositions.

Plaintiff’s response to the Reply? Nada.

After reviewing the Reply, the Court issued an order to show case, requiring Plaintiff’s attorney, Amy L.B. Ginsburg (“Ginsburg”), to appear in person with copies of the cases she cited.

Ginsburg’s response? Filed within an hour of the hearing.

In her filed response, Ginsburg “accept[ed] responsibility for the failure to verify the final version had been filed,” but then proceeded to place blame on her law clerk and paralegal. Specifically, Ginsburg represented that the law clerk drafted the Response which included citation of the fictitious case and quotes. In her review of the draft, Ginsburg allegedly recognized the fictitious case name as her husband had previously worked on a case with the same named plaintiff. Ginsburg supposedly recognized that the citation to the case was wrong and made a correction. She further claimed that she recently learned a paralegal mistakenly filed an earlier draft from the file which included the erroneous case citation.

In her filed response to the order to show cause, Ginsburg included the “correct” case. However, this “correct” case did not include the quotes cited in her Response. When asked about the quotations at the show cause hearing, Ginsburg simply argued that these quotes did not appear in her revised final draft.

Although her filed response to the order to show cause stated “it appears my law clerk did use AI to assist in drafting,” at the show cause hearing, Ginsburg said she could not know for certain if AI was used as the law clerk who drafted the original Response was no longer with her firm. She instead offered highlighted copies of the case she assumed the law clerk may have been referring to.

Ginsburg represented that she assumed the case was cited erroneously as it was a motion in limine order that would be more difficult to find in Lexis or Westlaw. The court simply wasn’t buying it and presented several issues with this argument. First, the court found that someone erroneously citing “the case name, number, year, and quotation…[is] not struggling with research or citation formatting. They are struggling with honesty.”

Ouch.

The court also had issue with Ginsburg’s “suspect” phrasing of the law clerk having difficulty locating the “correct case or cite,” noting that “[t]hese are not interchangeable details,” and that discovering a fictious case offered as a real one “would warrant a substantially different reaction than discovering, say, a typo in the Westlaw reporter number (i.e., discovering an incorrect cite).”

At the hearing, the Court brought to Ginsburg’s attention that other attorneys have attempted to avoid sanctions by offering similar excuses (e.g. paralegals filing wrong documents). Ginsburg explained that she had no idea since she had practiced over 20 years without anything like this happening, although she admitted to attending “numerous CLEs on the use of AI.”

After hearing Ginsburg’s explanations, the court found that her excuses “at best, lack credibility.” The court stated that although Ginsburg claimed to take ownership of the errors, she has not accepted responsibility, and her story was “a contrived attempt to shift the blame to others.” As such, the court determined that neither reprimand nor financial sanctions would be sufficient to deter similar misconduct.

In its attempt to order sufficient deterrents, the court considered the database of “AI Hallucinations” cases by lawyers in the United States cited by the Fifth Circuit (citing AI Hallucination Cases, Damien Charlotin, https://www.damiencharlotin.com/hallucinations). The number of cases in that database has almost doubled in just four months, growing from 239 cases in February 2026 to 434 cases in June 2026.

After considering Ginsburg’s “failed attempt to avoid responsibility” and the ever-growing cases of lawyers citing AI Hallucination cases, the court publicly reprimanded Ginsburg, ordering a copy of the Order to Sanction be submitted for publication in the Federal Supplement as well as to all district judges and magistrate judges in the district. Further, Ginsburg was ordered to verify the existence and accuracy of the legal authorities she cited in every filing this year. If any fictional authority or material misrepresentation is found, she has three days to notify the affected court and parties to correct the record.

The court didn’t stop there. On top of financial penalties, the court also ordered Ginsburg to attend CLE courses regarding the ethical use of AI in the practice of law and to include a signed Certification of Verification in every filing in which she cites any legal authority. The Certification of Verification must state the following:

“I, Amy L.B. Ginsburg, certify that I personally reviewed each citation and authority referenced in this filing and that, to the best of my knowledge and belief, all citations are genuine and accurately reflect the law as of the date of filing. I also certify that I personally verified that the file I reviewed is the same one that is being submitted to this Court.”

The lessons here are self-evident:

  • KEEP GENERATIVE AI OUT OF THE PRACTICE OF LAW.
  • Make sure you are performing a reasonable inquiry into cases and quotations cited in documents you are drafting or reviewing prior to filing.
  • Check filed documents to confirm accuracy.
  • Also perform a reasonable inquiry into the cases and quotations cited in filings by opposing parties.
  • Own your errors.
  • #DeserveToWin



Source link

Leave a Reply

Your email address will not be published. Required fields are marked *