Sanction teardown · W.D. Pennsylvania, USA · 2026-07-15
Mullins v. Duquesne University of the Holy Spirit
What happened
In W.D. Pennsylvania, USA, a filing relied on Claude to help draft legal argument. The court identified the following problems with the citations in that filing:
-
Misrepresented (Case Law)Defendants pointed to lengthy multi-case string citations and other citations that did not directly support adjacent propositions; Court found such citations confusing or inapt but not false or fabricated.
-
Misrepresented (Case Law)Defendants alleged plaintiff cited Phillips v. County of Allegheny for Rule 9(b)/fraud and document-handling doctrines that do not appear in Phillips; Court found Phillips did not address Rule 9(b) but plaintiff's meaning was decipherable and not sanctionable.
-
Misrepresented (Case Law)Defendants argued plaintiff cited Max's Seafood Café v. Quinteros for propositions not appearing in that decision and for challenging application of later cases; Court noted the citation did not support the specific phrases but was not a fabricated authority.
-
Misrepresented (Case Law)Defendants contended plaintiff misapplied Twombly and Pension Benefit Guaranty Corp. as addressing the standard for leave to amend; Court observed neither case addressed that specific standard but found the citations not to be non-existent or fabricated.
Which AI tool
Claude. Note: Charlotin's public database records tool attribution only where a court order, brief, or reporting on the matter states it explicitly; "unidentified" or "implied" means the record indicates AI use but does not name a specific product — we do not guess.
Outcome
Motion to strike denied
Additional detail
Defendants argued the pro se plaintiff's brief contained AI-generated inaccuracies, improper citations, and unintelligible phrasing suggesting reliance on generative AI. Plaintiff certified he used Anthropic's Claude for drafting and editing but not for generating citations or quotations. The Court reviewed cited examples (Phillips; Max's Seafood; Twombly; PBGC), found no fabricated or non-existent authorities and concluded the citations, while sometimes inapt or confusing, did not constitute sanctionable AI hallucinations. The motion to strike was denied and no sanctions were imposed.
How Citation Safe would have caught this
Citation Safe runs three deterministic layers before a brief is filed: (1) does the citation exist against CourtListener's database of published opinions, (2) if quoted, does that exact language appear in the source, (3) does the cited case actually support the proposition it is cited for. Fabricated case citations fail Layer 1. Fabricated or misattributed quotations fail Layer 2 even when the underlying case is real. Misrepresented holdings — a real case cited for a proposition it does not support — are the target of Layer 3. None of these checks involve asking another language model whether the citation looks right; they are lookups and text-matches against the actual source, which is why a hallucinated citation has to survive a direct lookup against the authoritative source — not another model's opinion — to earn a VERIFIED stamp; our measured false-verify rate is published live at /quality.
Check a brief before you file it → · See our live false-verify rate
Source: https://www.damiencharlotin.com/documents/2691/Mullins_v._Dusquesne_U._USA_15_July_2026.pdf, via Damien Charlotin's public AI Hallucination Cases Database (CC0).