A federal appeals court fines a lawyer $2,500 for an AI-drafted brief and evasive answers

February 2026 — A federal appeals court ordered a consumer-credit lawyer to pay $2,500 after finding she used AI to draft "a substantial portion, if not all, of her reply brief" and then did not check it. Here's what the court found, what it cost, and what went wrong.

Cost and outcome (as of Feb. 18, 2026)

Outcome: the court found counsel "used artificial intelligence to draft a substantial portion, if not all, of her reply brief and then failed to verify the accuracy of the content generated." (Fifth Circuit order)
Cost: a $2,500 sanction, payable to the clerk of court within 30 days. (Fifth Circuit order)
Lawyer's account: she said she used AI only to "help organize and structure [her] arguments and to break up overly long paragraphs," which the court did not find credible. (Fifth Circuit order)
Status: sanction ordered on Feb. 18, 2026, the date the order was filed. (Fifth Circuit order)

The case behind the brief

The appeal came out of a lawsuit by Robert Fletcher against Experian and Bridgecrest Credit Company under the Fair Credit Reporting Act, the order shows. A lower court had sanctioned Fletcher's original lawyer, Shawn Jaffer, and his firm; the Fifth Circuit had already ruled on that award in January, the order says.

"But a problem remained," wrote Chief Judge Jennifer Walker Elrod for a panel that included Judges Smith and Wilson. Heather Hersh, a member of the Jaffer & Associates firm, had filed a reply brief "containing numerous inaccurate citations, quotations, and statements of fact."

On Dec. 18, 2025, the court ordered her to show cause why she should not be sanctioned. The order listed "16 instances of fabricated quotations and 5 additional serious misrepresentations of law or fact." Hersh was the only attorney to sign the brief.

What the court checked

Hersh first said she relied on "publicly available versions of the cases" and that "certain paraphrased summaries did not match the reporter text," the order says.

The judges then went looking. Google Scholar, the first source she named, "does not contain summaries of legal cases that could contain inaccurate quotes," they wrote. They reviewed the CourtListener, Justia and FindLaw summaries for each case in the brief, and "the problematic quotations and propositions are nowhere to be found." The remaining two, Casetext, "which has been spun off into Thomson Reuters's CoCounsel," and vLex, "are both generative AI products," the court wrote.

Asked directly, Hersh "finally admitted to the use of AI," the order says. "Even when asked directly, Hersh did not explain the steps she took to verify the factual assertions in her brief, so the court concludes that she took none."

Why the penalty landed where it did

The court said it was "likely" she also used generative AI in her response to the show-cause order. "Had Hersh accepted responsibility and been more forthcoming, it is likely that the court would have imposed lesser sanctions," the panel wrote. "However, when confronted with a serious ethical misstep, Hersh misled, evaded, and violated her duties as an officer of this court."

The order also explains why the court has no AI-specific rule. In 2024 a three-judge subcommittee proposed requiring lawyers to certify either that no generative AI was used or that a human had checked the output. The court published it and received 13 comments; it then declined, concluding, "existing rules were sufficient to deter misconduct related to generative AI use."

The judges traced the problem back to Mata v. Avianca in 2023, where an AI-generated brief cited seven nonexistent cases, the order says.

What went wrong

No one checked the output. The brief went out with 16 fabricated quotations and five other serious misstatements, according to the court.
The first answer dodged the question. The court called her description of AI output as "paraphrased summaries" "evasive, misleading, and sanctionable."
Fabrications were subtler than fake cases. The court noted the trend toward invented quotations or statements of law "attributed to real cases, rather than the more easily recognizable fake cases."
What we don't know yet: which AI tool produced the text, whether the firm changed its review process, and whether any bar referral followed.

Our follow-up question for Jaffer & Associates

What review step does the firm now require before a brief drafted with AI is filed?

Sources

U.S. Court of Appeals for the Fifth Circuit, order, Fletcher v. Experian Information Solutions, Inc., No. 25-20086, Feb. 18, 2026. ca5.uscourts.gov
Reuters, “US appeals court orders lawyer to pay $2,500 over AI hallucinations in brief,” Feb. 18, 2026. reuters.com

Disclosure: [not in copy] Editor placeholder