Skip to content
Submit incident
AII-1001-37971
Under investigationA regulator, court or police has opened a formal inquiry and reached no finding yet.Supporting sources are pending verification.

Portuguese Judge Under Discipline for Using ChatGPT to Draft Court Ruling with Fabricated References

A Portuguese appellate court judge is under disciplinary investigation after allegations that ChatGPT was used to draft part of a ruling in a case involving a former member of parliament and a charity, with the text containing non-existent legal citations and language errors. The judge denies the allegations, and the investigation is ongoing with no finding yet reached.

3 October 2025 · Santa Casa da Misericórdia · Reliability and hallucination
3 October 2025
AII-1001-37971
Quick facts
Category
Reliability and hallucination
Country
Portugal
Date
3 October 2025
Drafted by
Raidu Incident Detection

What happened

A judge at Tribunal da Relação de Lisboa allegedly used ChatGPT to draft part of a court ruling in a case involving former member of parliament Helena Lopes da Costa and Santa Casa da Misericórdia.

Defence lawyers detected unusual expressions, Portuguese language errors, and legal references that did not correspond to applicable legislation, including a citation to 'article 22 of Law 34/87 of 16 July' to support a finding of abuse of power, which does not exist in the cited law.

The judge denied using any artificial intelligence tool and claimed to have written the ruling himself. The Conselho Superior da Magistratura opened a disciplinary process to determine responsibility.

Reported impact

The inconsistencies raised serious doubts about the integrity and validity of the court decision.

Index analysis

Twelve defence lawyers filed a complaint with the Conselho Superior da Magistratura demanding an investigation into the alleged use of AI. The Tribunal da Relação de Lisboa initially dismissed the allegations as 'completely unfounded', then corrected the ruling but without altering the decision, after which the defence lawyers stated their suspicions had strengthened rather than been resolved.

Governance controls

The Conselho Superior da Magistratura confirmed that its plenary received the report of preliminary investigation and decided to open a disciplinary process, which is conducted confidentially.

Classification

Category
Reliability and hallucination
Industry
Government, security and defence
Country
Portugal
Harm status
Potential harm
Intent
Unknown
Timing
Post-deployment
Harm types
Public interest, Reputational
Failure modes
Fabricated citation, Hallucination
Consequences
Regulatory investigation
AI systems
ChatGPT

Sources and evidence

Sources not yet published

This record was researched and written by the Index from original reporting. Its sources are still being verified by an editor, and they are listed here once that check is complete. Until then the record carries no reviewer.

How sources are verifiedSuggest a source for this incident