Dodds v. Bridges
U.S. Court of Appeals for the Tenth Circuit · 10th Cir. · Colorado bar guidance , Kansas bar guidance , New Mexico bar guidance , Oklahoma bar guidance , Utah bar guidance , Wyoming bar guidance
Conduct
Pro se § 2254 appellant cited three reporter-mismatched cases (US v. Farrow, US v. Matthews, US v. Mateo) plus fabricated quotations.
Consequence
§ 2254 denial affirmed. Stand-alone warning to Dodds and all 10th Cir. pro se litigants and counsel; sanctions possible next time.
Lesson
10th Cir. treats reporter-cite mismatches plus fabricated quotes as AI-hallucination, even when the litigant denies AI use.
Verified May 8, 2026
- Citation
- Dodds v. Bridges, No. 25-7021 (10th Cir. Feb. 11, 2026) (Carson, Baldock, Kelly, JJ., op. by Baldock, J.) (unpublished)
- Decided
- February 11, 2026
Summary
Pro se Oklahoma state prisoner Thomas Carl Dodds, Jr., appealed the district court's denial of his 28 U.S.C. § 2254 habeas petition, challenging a sentence-related due process claim under Tucker. The Tenth Circuit panel (Carson, Baldock, Kelly) identified three citations in Dodds's appellate briefing where the reporter cite pointed to an unrelated case rather than the cited authority: United States v. Farrow, 599 F.2d 154 (citation actually leads to Illinois v. City of Milwaukee); United States v. Matthews, 7 F.3d 1552 (citation leads to Post v. City of Fort Lauderdale); and United States v. Mateo, 476 F.3d 179 (citation leads to In re O'Lexa and Wishkin v. Potter). Quoted material attributed to Farrow and Matthews was likewise fabricated. The court attributed the conduct to unverified use of a generative AI tool "such as ChatGPT," even though Dodds claimed the brief had been typed by an unnamed outside source who introduced "transcription errors."
- AI tool:
- Generative AI inferred (panel: 'such as ChatGPT'; Dodds attributed the citations to 'transcription errors' by an unnamed outside source who typed the brief)
What sanction did the court impose?
Denial of habeas relief affirmed on the merits. The court declined to sanction or dismiss, instead issuing a stand-alone written warning to Dodds "and all pro se litigants and counsel appearing before this court" that fabricated citations may draw sanctions or dismissal in future filings. Motion to appoint appellate counsel was denied in the same order.
Why does Dodds v. Bridges matter for law firms using AI?
Dodds is the Tenth Circuit’s clearest published example of the reporter-cite-mismatch failure mode: a citation that looks valid on its face but points to a different case in the same reporter volume than the case named in the citation. Each of the three citations Judge Baldock identified followed the pattern, with Farrow’s reporter slot occupied by Illinois v. City of Milwaukee, Matthews’s slot by Post v. City of Fort Lauderdale, and Mateo’s slot by two unrelated decisions. The panel treats the pattern as an AI fingerprint: the kind of error that arises when a generative model produces a plausible reporter citation by predicting reporter volume and page numbers without checking the underlying case identity. Dodds’s defense, that an unnamed outside source typed the brief and introduced “transcription errors,” does not defeat the finding; the Tenth Circuit attributes the conduct to AI based on the citation pattern itself, not on the litigant’s account of how the brief was prepared.
For firms representing institutional defendants in 10th Cir. habeas appeals, the operational point is that a Dodds-style warning is now likely the first chambers response to an AI-fabricated brief from a pro se prisoner. The Carson/Baldock/Kelly panel has now issued three AI-warning orders in a four-month span (Chumpitaz-Morales and Biglow are the others), and the warning language explicitly extends to “all pro se litigants and counsel appearing before this court,” meaning a firm’s own represented filings inherit the same standard. The Wadsworth v. Walmart Inc., 348 F.R.D. 489 (D. Wyo. 2025) authority Baldock cites across the chambers cluster is the regional anchor for the “AI-resources-generate-fake-cases” proposition; firms preparing 10th Cir. briefs should expect Wadsworth to appear in any future AI-citation order from this circuit.
Implications for your firm
Operational steps a firm reading this case may wish to consider documenting. Strategic and rule-application calls belong to your firm's attorneys.
- When opposing pro se § 2254 or § 1983 prisoner appeals in the 10th Cir., flag reporter-cite mismatches and the underlying mis-cited cases on the record; the panel will name the failure mode and treat it as warning-bait without a separate motion.
- Cite Dodds for the proposition that a litigant's denial of AI use does not defeat the inferred AI-hallucination finding; the Tenth Circuit will attribute the conduct to ChatGPT-style tools based on the citation pattern itself, not the litigant's representation.
- The Carson/Baldock/Kelly chambers cluster has now issued three AI-warning orders in four months (Dodds, Chumpitaz-Morales, Biglow); track this panel composition for AI-citation matters in the 10th Cir.
Sources
Primary sources
- No Westlaw assignment is visible in the unpublished opinion's PDF header at this verification pass.