Markwayne Mullin immigration appeal draws AI admonishment

Risk Digest entry — not legal advice. For readers arriving through “markwayne mullin immigration law backlash news,” the sourceable court event is narrower than the politics around the name: Dec v. Mullin, No. 25-2417, a Seventh Circuit immigration appeal naming DHS Secretary Markwayne Mullin, in which the panel affirmed a jurisdictional dismissal and separately admonished counsel over two nonexistent case citations in the appellate brief.[1][2]

What the Seventh Circuit actually found

The useful part of Dec v. Mullin is not that an immigration brief may have involved generative AI. That is now too common a suspicion to do much work by itself. The useful part is that the Seventh Circuit identified a concrete defect: two cited cases did not exist, and the citations carried what the court described as “tell-tale signs of AI hallucinations.”[2]

That matters because a fabricated citation is not confined to the lawyer who filed it. Someone has to check it. A clerk may spend time looking for a case that is not there. The panel has to decide whether the error is isolated, reckless, or worse. Opposing counsel may have to respond to a phantom authority instead of the actual legal issue. In an immigration appeal, the client also inherits a credibility problem that has nothing to do with the visa dispute.

The court’s remedy was limited. Counsel was admonished, and the panel imposed no further sanction.[2] That is a meaningful outcome, but it is not an exoneration of the filing practice. An admonishment-only order still records a candor failure in a published appellate context, and the absence of a monetary sanction does not convert nonexistent citations into a harmless formatting mistake.

Separate the visa appeal from the citation problem

The underlying immigration ruling and the AI-citation issue should not be collapsed into one story. On the merits, the Seventh Circuit affirmed dismissal for lack of subject-matter jurisdiction in a family-based visa appeal.[1] The court’s disposition of jurisdiction did not depend on treating the fabricated citations as the reason the immigration claim failed.

For filing teams, that separation is the point. The client’s appeal already had a jurisdictional problem. The brief then added an avoidable reliability problem. Even where the sanction is only an admonishment, the attorney has forced the court to address counsel’s citation hygiene alongside the client’s immigration posture.

Nor does the record support tool speculation. The available account does not name the AI system allegedly used, and the court’s concern was not vendor attribution. The defect was the presence of nonexistent legal authorities in a federal appellate brief and the resulting candor issue.[2]

Why the sanction comparison matters

The Seventh Circuit outcome should be read beside, not merged with, the Ninth Circuit’s June 3, 2026 sanctions against immigration plaintiffs’ attorneys. In that reported decision, the Ninth Circuit imposed $2,500 sanctions on each attorney, six-month suspensions, and a two-year condition requiring disclosure of generative-AI use.[3]

That comparison is narrow but important. It does not prove that every circuit will punish AI-invented citations harshly. It also does not prove that fabricated citations usually end with only a warning. It shows that, in the same 2026 immigration-filing window, appellate courts treated the same broad category of failure with materially different remedies: admonishment and no further sanction in Dec v. Mullin, versus money, suspension, and disclosure conditions in the Ninth Circuit example.[2][3]

The filing signal for immigration and appellate teams

The immediate operational lesson is not “never use AI.” The Seventh Circuit order is more specific than that. The court was faced with a brief containing two authorities that did not exist, and the available public record does not identify a disclosed tool.[2] For a risk editor, that combination is worse than mere technology use: nonexistent citations, no tool clarity, and a court left to characterize the problem after filing.

The practical exposure also sits at the appellate level, where citation defects travel quickly. A district-court filing error can be serious; an appellate order naming the defect becomes briefing material for risk committees, malpractice reviews, opposing counsel, and judges in later cases. Dec v. Mullin is now one of those compact records: two fake citations, AI-hallucination language, admonishment, no further sanction.

For immigration lawyers, the setting is not incidental. High-volume filing conditions do not erase duties of candor, but they do explain why citation-verification workflows have to be designed for rushed briefs rather than ideal ones. The check that matters is not a vague instruction to “verify everything.” It is a process that makes it hard for a nonexistent case to survive into the filed PDF: source retrieval, reporter and docket confirmation, quotation matching, and final authority review by someone who is not simply trusting generated text.

A related citation-verification workflow belongs upstream of filing, not after a show-cause order or an appellate footnote. By the time a panel is identifying hallucinated authorities, the quality-control failure has already been exported to the court.

What this record does—and does not—support

The supported conclusion is jurisdiction-sensitive. Dec v. Mullin is a Seventh Circuit benchmark for admonishment-only treatment of AI-invented citations in an immigration appeal. The Ninth Circuit report shows that comparable immigration-filing failures can trigger monetary sanctions, suspensions, and disclosure obligations elsewhere.[2][3]

The record does not support a broad claim that federal courts have adopted a uniform AI-sanctions standard. It also does not support the softer claim that fabricated citations are usually forgiven if counsel blames a tool. The better reading is narrower and more useful: fabricated citations are now a documented candor problem in immigration appeals, and the penalty range depends heavily on the court applying it.

References

1. Dec v. Mullin, No. 25-2417, Justia, March 30, 2026
2. Attorney Escapes With Admonishment After 7th Circuit Fake Citations, Mealey’s/Law360
3. Ninth Circuit June 3, 2026 sanctions of immigration attorneys over AI “hallucinations” and lack of candor, Reuters, June 3, 2026