Sanction teardown · TJ San Catarina, Brazil · 2025-04-16
Nº 5025505-88.2025.8.24.0000/SC
What happened
In TJ San Catarina, Brazil, a filing relied on an unnamed/unconfirmed AI tool to help draft legal argument. The court identified the following problems with the citations in that filing:
-
Misrepresented (Case Law)Petition misstates the holding of RE 1.235.340 (STF), presenting the precedent as supporting the opposite thesis regarding execution after jury verdict.
-
False Quotes (Case Law)Excerpts attributed to HC 559.191/PR (STJ) do not reflect what that decision actually decided; quoted passages were created/altered by the AI.
-
Misrepresented (Case Law)HC 474.832 was cited with an incorrect state of origin (presented as Rio Grande do Sul though it is from the Distrito Federal) and used out of context.
-
Misrepresented (Case Law)HC 551.268 was cited with an incorrect state of origin (presented as Paraná though it is from São Paulo) and its excerpts do not match the actual decision.
-
Misrepresented (Case Law)Local TJSC process 5072992-59.2022.8.24.0000 was cited as a habeas corpus decision though the file is a criminal review (revisão criminal), mischaracterizing the precedent.
Which AI tool
an unnamed/unconfirmed AI tool. 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
Habeas Corpus request dismissed; Bar Referral
Additional detail
The court found the habeas corpus petition was drafted with clear use of artificial intelligence and contained multiple incorrect or fabricated citations. The petition mischaracterized an STF precedent (RE 1.235.340) and presented several STJ judgments with wrong origins and altered excerpts; one STJ number corresponded to a real process but its quoted passages were false. The tribunal concluded the AI-generated content was misleading and disrespectful, denied the relief, and ordered the file sent to the OAB Ethics Committee for evaluation of the lawyer's conduct.
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/2859/Jurisprud%C3%AAncia_Catarinense_-_TJSC_-_1.pdf, via Damien Charlotin's public AI Hallucination Cases Database (CC0).