Two U.S. Justice Department lawyers violated professional conduct rules in their use of AI — one submitted a court filing with fabricated quotations and then misled a judge about it, and the other uploaded work documents to a public AI platform despite court-ordered restrictions — according to summaries issued Tuesday by the department’s own ethics office. Why it matters: this is the Justice Department policing AI misuse inside its own ranks, and both cases have been referred to the unit that can discipline the lawyers.
November 18, 2026
What happened
The Justice Department’s Office of Professional Responsibility issued summaries of the two cases on Tuesday, Reuters reported. The summaries did not name the lawyers, say where they worked, or say whether they are still employed by the department. In both cases, the office’s findings were referred to the department’s Professional Misconduct Review Unit, which has the power to discipline misconduct. The Justice Department said in a statement that it “takes seriously its responsibility to review and address allegations of professional misconduct through appropriate established procedures,” and declined to provide further details.
Case one: fabricated quotations, then a misleading explanation
In the first matter, an assistant U.S. attorney submitted a court filing that, in the department’s words, “misrepresented case holdings and included fabricated quotations.” Pressed by the judge to explain the errors, the lawyer did not initially disclose his use of AI and instead told the court he had mistakenly filed a draft of the filing. The ethics office found the prosecutor made false and misleading statements about his use of generative AI to the judge at a hearing. The AI platform involved was not named.
Case two: restricted documents uploaded to a public AI platform
In the second matter, an attorney uploaded work-related documents to a public AI platform despite court-ordered restrictions on how the information could be shared. The ethics office said the attorney “recklessly violated the rules of professional conduct when he used the public platform in violation of department policy.” Reuters noted it was not immediately clear whether the findings were also referred to state agencies that oversee attorney conduct.
Why it matters
AI-generated errors in court filings have been drawing sanctions for years, but these findings land differently: they come from the Justice Department’s own ethics office, about its own lawyers, and they cover the two failure modes every legal team should worry about — invented citations presented as real, and confidential material pasted into a public tool. Reuters framed the cases as a rare window into how the department is policing generative AI as it spreads through the legal profession. The context is not isolated, either: in May, a North Carolina state bar grievance committee publicly reprimanded a federal prosecutor who had left his post after submitting a brief in a veterans’ benefits case that contained fabricated quotes and other errors. For any professional using AI on real work, the lesson is plain and practical: verify every citation yourself, and never put restricted documents into a public AI service. Everything AI, tested — including, apparently, inside the Justice Department.
Frequently asked questions
What did the two Justice Department lawyers do?
According to the department’s Office of Professional Responsibility, one assistant U.S. attorney filed court papers that misrepresented case holdings and included fabricated quotations, then made false and misleading statements to a judge about his AI use. Another attorney uploaded work documents to a public AI platform despite court-ordered restrictions.
What happens to the lawyers now?
Both findings were referred to the Justice Department’s Professional Misconduct Review Unit, which has the power to discipline misconduct. The summaries did not name the lawyers or say whether they still work at the department.
Which AI tool was involved?
The summaries did not name the AI platform in either case.
Sources: Reuters

