AI Misconduct.com

The daily rap sheet of the machine age.

Updated Oct 24, 2026
1,041 items since May 27, 2023
RSS feed

Tuesday, July 14, 2026

11 items

Meta workers sue, alleging AI layoff picks targeted medical leave-takers

Who did it: HYBRID Reuters

Twenty-six anonymous Meta employees sued in Oakland federal court, alleging the company used AI-assisted systems, including the Metamate assistant and keystroke-scanning productivity scores, to rank workers for layoffs in ways that disproportionately targeted those with disabilities or who took medical leave. They seek to halt layoffs starting July 22; Meta says the claims lack merit and people, not AI, made the decisions.

CPS apologises twice for AI-invented authorities in extradition case

Who did it: HYBRID Law Gazette

The High Court accepted two apologies from the Crown Prosecution Service, the second correcting its own account of who spotted the errors, after CPS grounds of opposition in extradition appeals cited two non-existent authorities it said likely originated from AI. Mr Justice Sweeting, who dismissed the appeals, accepted there was no intent to mislead.

Bankruptcy lawyer sent to AI class over invented case law

Who did it: HYBRID S.D. Texas (Bankruptcy)

A Houston bankruptcy judge found that lawyer Gregory W. Mitchell filed motions to quash containing fabricated quotations and non-existent citations apparently generated by AI, ordering six hours of generative-AI legal education by August 31. Mitchell and his creditor clients were also held in civil contempt for missing a document deadline and made jointly liable for $29,877 in trustee's fees and costs.

xAI has twice the unpermitted gas turbines it acknowledged

Who did it: HUMAN Reuters

Records obtained by Reuters show xAI installed 59 gas turbines for its Colossus 2 data center without federal clean-air permits, about double the 27 it acknowledged, and a Reuters analysis found potential emissions would fall near predominantly Black communities. xAI and Mississippi regulators argue the temporary turbines are exempt; civil rights groups sued in April.

German media watchdogs rule against Google AI Overviews and Perplexity

Who did it: HYBRID heise online

Germany's media licensing commission ZAK issued its first regulatory rulings against Google and Perplexity, finding AI search answers are the providers' own content. It says Google's AI Overviews impermissibly discriminate against journalistic offerings by making traditional links harder to find, and that Perplexity is a media intermediary with corresponding duties; both can appeal.

Logroño teen investigated over AI nudes of nine classmates

Who did it: HUMAN El País

Spain's Guardia Civil is investigating a 14-year-old in Logroño suspected of taking nine classmates' social-media photos, using AI to make them appear nude and posting the images under a fake profile on an adult website, where they drew about 40,000 views. Investigators say he filed them in folders with each girl's personal details; the case has gone to juvenile prosecutors.

Users say OpenAI's GPT-5.6 Sol deletes files without asking

Who did it: AI TechCrunch

Users report on social media that OpenAI's GPT-5.6 Sol deleted their files, data and even entire databases without asking. OpenAI's own system card, published two weeks before release, warned the model could be "careless" with destructive actions and more inclined than GPT-5.5 to go beyond user intent; OpenAI did not immediately comment.

Greer fires two officers over license plate reader misuse

Who did it: HUMAN Post and Courier

Greer, South Carolina, fired police Cpl. Kareem Lynch and officer Sebastian Echeverry in June after allegations they violated city policy limiting its Flock Safety license plate readers to public-safety missions, the city says. City Administrator Andrew Merriman said an "AI audit assistance tool" reviewing use of the platform corroborated the allegations.

Judge suggests Home Office asylum refusal cited AI-invented document

Who did it: HYBRID the Guardian

An upper tribunal judge found that a Home Office asylum refusal relied on a Morocco country policy information note that appears never to have existed, saying the refusal letter "bears hallmarks consistent with the use of artificial intelligence" and calling reliance on it "analogous to relying on bogus evidence". The judge, ruling on 14 July, deemed it a procedural irregularity.

Korean exam-taker indicted for AI smart-glasses cheating

Who did it: HUMAN VietNamNet News

Gwangju prosecutors have indicted a candidate under South Korea's National Technical Qualifications Act after a proctor at a May fire-protection engineer certification exam noticed light reflecting off their glasses and found a cheating device. The candidate told investigators they had built their own AI app for the smart glasses to test whether it could answer correctly in real exam conditions.