ECLI:NL:RBGEL:2025:9423

ChatGPT (suspected) fake case citationwrong citation to real sourceno human verificationsuspected generative ai hallucination

I. Executive Summary

One day before the hearing, the claimant’s representative cited numerous purported case-law references, including ECLI numbers. The court verified that many references were inconsistent with the cited dates and ECLI numbers, irrelevant, or did not exist, and the representative could not clarify their existence or relevance. The court stated it could not escape the impression that ChatGPT or another AI tool was used. Because some decisions did not exist and relevance was unclear, the court disregarded the cited ‘case law’.

II. Conduct Analysis

A party’s representative submitted multiple purported case-law citations that were mismatched, irrelevant, or non-existent; the court suspected use of ChatGPT or another AI tool and found the citations unverifiable.

III. Legal Foundations

Applicable procedural rules and judicial ethics codes.

IV. Key Facts

1) The representative submitted additional grounds citing many purported decisions with ECLI numbers shortly before the hearing. 2) The court could find only two relevant decisions; other citations did not match the stated date/ECLI combination or did not exist, and some ECLI numbers related to unrelated topics. 3) The representative could not explain existence, location, or relevance; the court suspected ChatGPT or another AI tool and disregarded the citations.

V. Consequences & Sanction

1) The court disregarded the alleged AI-linked case-law citations. 2) No sanction proceeding or procedural penalty is described in the decision for the citation issue.