Sanction teardown · ICSID Tribunal, International Arbitration · 2025-08-26
Orano Mining v. Niger (2)
What happened
In ICSID Tribunal, International Arbitration, a filing relied on Implied (by me) to help draft legal argument. The court identified the following problems with the citations in that filing:
-
Fabricated (Case Law)Citing a disqualification decision from Antin v. Spain that does not exist
-
Fabricated (Case Law)Citing a decision on disqualification from Crystallex v. Venezuela that does not exist
-
Fabricated (Case Law)Citing a decision on challenge from Repsol v. Argentina that does not exist
-
Fabricated (Case Law)Citing a decision on challenge from Transglobal v. Panama that does not exist
-
Fabricated (Case Law)Citing a decision on disqualification from Rusoro v. Venezuela that does not exist
-
Misrepresented (Case Law)Misrepresenting a paragraph in a decision in Perenco v. Ecuador as bearing on conflict on interests
-
False Quotes (Case Law)Citing from para. 91 of a decision in Abaclat v. Argentina that contains only 83 paragraphs
Which AI tool
Implied (by me). 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
Arguments ignored
Additional detail
Although the tribunal did not address it as a case of hallucinations or misuse of artificial intelligence, the details make clear that this was very likely at issue. While the decision is not public, further details have been reported by, uh, me:🔗 Damien Charlotin, Niger’s proposal to disqualify Fernando Mantilla-Serrano from uranium mining arbitration is rejected; challenge procedure is marred by citations and authorities that couldn’t be borne out when scrutinized by co-arbitrators (Investment Arbitration Reporter, 28 August 2025).
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: Damien Charlotin's public AI Hallucination Cases Database (CC0).