Sanction teardown · NSA, Poland · 2026-08-27
I OSK 935/26
What happened
In NSA, Poland, 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 (Legal Norm)The appeal attributed the principle of two-instance administrative proceedings to Article 35 of the Code of Administrative Procedure, although Article 35 concerns statutory deadlines for resolving administrative matters; the principle is provided by Article 15.
-
Misrepresented (Legal Norm)The appeal claimed that Article 87 § 1 of the Code of Administrative Procedure permitted representation by an appointed attorney despite the party's death. Article 87 has no paragraphs and concerns delegation of evidentiary proceedings by a collegiate administrative body.
-
Misrepresented (Legal Norm)The appeal invoked Article 12 of the Civil Code as establishing a fundamental principle of respect for a deceased person's rights and legal succession. The NSA noted that Article 12 instead concerns the lack of capacity to perform legal acts of young or fully incapacitated persons.
-
Fabricated (Legal Norm)The appeal referred to an alleged 17 June 2004 Act on proceedings concerning social assistance, which the NSA stated does not exist in the Polish legal system.
-
Misrepresented (Doctrinal Work)The appeal asserted that the first-instance court had violated the two-instance principle by citing Article 35 of the Code of Administrative Procedure, and used that erroneous citation as part of its cassation arguments. The NSA treated the claims as unsupported and outside the properly formulated grounds of cassation.
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
The NSA dismissed the cassation appeal and criticized the professional lawyer's unverified AI-generated arguments as gross negligence; no separate sanction was imposed.
Additional detail
The NSA identified several apparently AI-generated errors in the cassation appeal: Article 35 of the Code of Administrative Procedure was cited as establishing the two-instance principle instead of governing decision deadlines; Article 87 of that Code was wrongly described as concerning representation after a party's death; Article 12 of the Civil Code was falsely attributed a principle concerning respect for deceased persons and succession; and the appeal invoked a nonexistent 17 June 2004 statute on social-assistance proceedings. The court held that these were not merely drafting errors, regarded the lawyer's failure to verify the AI-generated content as gross negligence, and dismissed the appeal on the merits because it lacked valid grounds.
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/3227/I_OSK_935_26_-_Postanowienie_NSA_z_2026-08-27.pdf, via Damien Charlotin's public AI Hallucination Cases Database (CC0).