AI Misconduct.com

The daily rap sheet of the machine age.

Updated Nov 2, 2026
4,173 items in the archive
RSS feed

Friday, October 3, 2025

8 items

Alberta lawyer who banned AI held responsible for contractor's hallucinations

Who did it: HYBRID Canadian Lawyer

The Alberta Court of Appeal ruled that Calgary lawyer Christopher Souster "bears ultimate responsibility" for a factum citing hallucinated case law, though it was drafted by a contractor who assured him no large language model was used. Souster had banned generative AI at his firm since November 2024; the court let him file an amended factum.

Italy's privacy watchdog blocks Clothoff nudify app

Who did it: HUMAN ANSA

Italy's data protection authority ordered an urgent, immediate provisional limitation on processing Italian users' data by the British Virgin Islands company behind Clothoff, an app that generates fake nude images and videos of real people. The Garante said anyone, minors included, could use it, with no consent check or label marking content as artificial, and opened an investigation into all nudify apps.

Lisbon appeals judge faces disciplinary case over alleged AI use

Who did it: HYBRID Lusa

Portugal's Superior Council of the Judiciary opened disciplinary proceedings against a Lisbon Court of Appeal judge accused of using AI to draft a ruling that cited non-existent legislation and case law. Twelve defence lawyers complained; the court first called the AI allegations 'completely unfounded', then corrected the ruling without changing the decision.

TikTok steered test child accounts to porn, campaigners say

Who did it: AI The Guardian

Global Witness says TikTok suggested sexualised search terms to seven test accounts registered as 13-year-olds with 'restricted mode' on, leading to pornographic content within a small number of clicks, in one case two. The group claims this breaches the UK's Online Safety Act; TikTok said it removed the videos and improved its search suggestions.

Suno disputes labels' new YouTube stream-ripping claims in AI suit

Who did it: HUMAN Billboard

On Oct. 3 Suno asked a judge to reject Universal, Warner and Sony's bid to add claims that it violated the DMCA by stream-ripping songs from YouTube to train its models; the labels seek $2,500 per violation. Suno argues the law bars circumventing access controls, not copying freely available material: 'Anyone can access YouTube content.'

Northern Data probed over €100M VAT break on Nvidia chips

Who did it: HUMAN Wccftech

Swedish authorities are investigating Tether-backed AI cloud firm Northern Data over a tax break of approximately €100 million on Nvidia H100 chips worth around €400 million, amid allegations the chips were used for cryptocurrency mining rather than AI work. According to Bloomberg, raids in Frankfurt and Boden, Sweden, led to arrests.

D.C. judge sanctions lawyer whose AI-researched brief botched every citation

Who did it: HYBRID AI Hallucination Cases database

Judge Amit P. Mehta ordered lawyer Glenn Ellis to pay the defendant's fees and costs for two filings after Ellis's brief, built with an AI legal research tool whose output he did not double-check, cited nine cases: four for false propositions, four with fabricated quotations and one nonexistent. The court called the conduct 'reckless' and 'singularly egregious'.

Lawyer blames wrong PDF for AI-invented case citations

Who did it: HYBRID AI Hallucination Cases database

Judge Louise W. Flanagan found that a sentencing memorandum filed by defence lawyer Lauren H. Brennan cited nonexistent cases, cases from the wrong circuit and quotations the cases do not contain. Brennan, raising the errors herself, said she used AI to frame her argument and mistakenly filed her first draft; the court let her file a corrected brief, noting 'there are lessons to be learned'.