Doe v. United States (D.N.J. 2026)

What sanction did the court impose?

Judge Christine P. O'Hearn granted Plaintiff's IFP application but dismissed the complaint without prejudice under 28 U.S.C. § 1915(e)(2) for failure to state a claim on which relief may be granted, denied the TRO/PI motion without prejudice, and granted 30 days to amend. The AI conduct was addressed in footnote 3 rather than as a stand-alone holding: 'Some of Plaintiff's filings seem to have been generated using artificial intelligence ('AI') and include what appear to be hallucinated or nonexistent legal authorities. (See Compl. Ex. 1, ECF No. 1-3 at 1). Although pro se pleadings are construed liberally, an 'unrepresented party' remains subject to Federal Rule of Civil Procedure 11 and may be sanctioned for failing to conduct a reasonable inquiry into whether the claims, defenses, and other legal contentions are warranted by existing law. Thus, to the extent Plaintiff submits any additional filings, she is cautioned as to her obligations under Rule 11.' The Court cited Powhatan Cnty. Sch. Bd. v. Skinger, No. 24-874, 2025 WL 1559593, at *10 (E.D. Va. June 2, 2025) (collecting cases) as the cross-circuit authority for the proposition that courts routinely 'threaten to impose sanctions on litigants proceeding pro se who cite AI-hallucinated or otherwise nonexistent legal authority to support their positions.' The Court also separately dismissed claims brought on behalf of Plaintiff's minor child, citing the Third Circuit rule that a parent appearing pro se may not bring an action on behalf of their minor children (Osei-Afriyie v. Med. Coll. of Pa., 937 F.2d 876, 882-83 (3d Cir. 1991)).

Why does Doe v. United States (D.N.J. 2026) matter for law firms using AI?

The structural choice in O’Hearn’s order is worth noting for partners studying the 2026 corpus. The Court did not need to reach the AI conduct to dismiss; Rule 8 vagueness and § 1915(e)(2) failure-to-state-a-claim were independently sufficient. Adding the AI warning in footnote 3 puts the issue on the docket without making it the operative basis for the dismissal, which sidesteps any later argument that Plaintiff’s substantive claims were prejudiced by the AI issue. The footnote is also tied directly to the leave-to-amend grant: the warning is functionally a “if you amend, do not repeat this conduct” note, which gives the Court a docketed predicate for sanctions on any post-amendment repeat.

The Powhatan v. Skinger cite carries weight in this corpus. That case is itself a Schiller-chambers (E.D. Va.) collecting-cases opinion on pro se AI sanctions; it is increasingly the authority sister courts cite when they want to anchor an AI-caution footnote in a cross-circuit consensus rather than a single-judge view. For partners drafting AI-citation sanctions briefs in 2026, Powhatan is the working canonical cite for “courts have routinely threatened to impose sanctions on pro se litigants who cite AI-hallucinated authority.”

Finally, the procedural shape (warning in footnote, dismissal on other grounds, leave to amend) is now visible across multiple districts in early 2026: Ambrose v. Lee (D. Conn., footnote-warning to pro se defendant), Dixon v. MultiCare (W.D. Wash., warning in section G of an order on motion to dismiss), Doe v. United States (D.N.J., footnote 3 on a § 1915 dismissal), and CHP 1010 McDowell v. Turpen (Bankr. D. Colo., separately-labeled “Warning” section in an AP motion to dismiss order). The judicial response is calibrating to: dismiss / decide on independent grounds; add the AI warning at the procedural moment closest to the next-filing decision; reserve fees for any repeat conduct.