What was hallucinated

Fabricated: Case Law | Appellate brief cited multiple purportedly binding decisions (attributed to this Relator, to other TJSP panels, STF and STJ) that could not be located and were found to be AI-generated. || Misrepresented: Exhibits & Submissions | The petition itself showed signs of having been produced by a generative AI (retained template text like "[indicar páginas do interrogatório]").

Details

The court found that the defense's brief contained jurisprudence that did not exist and that the cited 'decisions' were generated by a generative AI tool (no specific tool named). The court relied on formatting, absence of relator/organ/judge/process numbers, and presence of template placeholders (e.g. "[indicar páginas do interrogatório]") to conclude the pieces were AI-produced. It held this conduct as abusive and constituting litigância de má-fé, applied formal censure and a fine (amount unspecified), and emphasized the duty to verify sources and the inadmissibility of relying on AI-generated, unverifiable jurisprudence.

Sanction teardown · TJ San Paulo, Brazil · 2026-03-10

Nº 1500026-62.2023.8.26.0633/50000

What happened

In TJ San Paulo, 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:

Fabricated (Case Law)

Appellate brief cited multiple purportedly binding decisions (attributed to this Relator, to other TJSP panels, STF and STJ) that could not be located and were found to be AI-generated.
Misrepresented (Exhibits & Submissions)

The petition itself showed signs of having been produced by a generative AI (retained template text like "[indicar páginas do interrogatório]").

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

Formal Reprimand; Monetary Fine (monetary penalty: 1 BRL.)

Additional detail

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/2862/20260000198236.pdf, via Damien Charlotin's public AI Hallucination Cases Database (CC0).

Don’t be the next case in this database.

Citation Safe checks every citation against primary sources before it reaches a filing.

Try the free checker Back to database

On the opposing side of their next motion?

Check their next brief through the same engine that would have caught this filing — $49, no signup, full annotated report you can attach to your Rule 11 or sanctions motion.

Open Opposition Check — $49 Sanctions-motion template