Sanction teardown · T.S.X. Galicia, Spain · 2026-07-07
Recurso de Suplicación 0005472/2025
What happened
In T.S.X. Galicia, Spain, 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)Court found quoted material attributed to SSTC 60/2008 was fabricated and not present in that ruling.
-
Fabricated (Case Law)Passages attributed to STS 459/2022 were identified as invented and do not correspond to any paragraph of that decision.
-
Fabricated (Case Law)Citation STS 23/03/21 R. 2341/19 was found not to exist (apocryphal resource number) though quoted in the brief.
-
Fabricated (Case Law)Court identified the quoted excerpt attributed to STS 12/04/19 R. 4123/17 as a false quotation (does not match the ATS referenced).
-
Fabricated (Case Law)The brief attributed a passage to the STJUE C-127/2019 that does not appear in that preliminary‑reference file; court found the quotation invented.
-
Fabricated (Case Law)Court determined quoted passages attributed to SSTC 142/2015 were invented and do not appear in that decision.
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
Bar Referral (monetary penalty: 1800 EUR.)
Additional detail
Earlier decision noticing the issues here.In an innovative approach, the court set the fine amount by reference to the annual cost of a verified legal AI tool (citing products such as Sof-IA by Tirant lo Blanch, GenIA-L by Lefebvre, Harvey AI, and others), reasoning that the use of such a tool would have eliminated or drastically reduced the risk of hallucination
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/2614/Galicia_7_July_2026.pdf, via Damien Charlotin's public AI Hallucination Cases Database (CC0).