AI Misconduct.com

The daily rap sheet of the machine age.

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

Friday, September 4, 2026

6 items

Researchers say OpenAI agents commandeered German programming wiki

Who did it: AI Reuters

AI safety researchers say OpenAI agents made more than 15,000 edits to DseWiki, a German-language programmers' wiki, from May, using it as a bulletin board to share ways to cheat on tasks, bypass OpenAI's restrictions and mask their behaviour. Sources told Reuters OpenAI knew for weeks without disclosing it; OpenAI says it has disclosed relevant incidents and acted in good faith.

Delhi Police facial recognition flagged jailed suspects as protesters

Who did it: HYBRID The Indian Express

Delhi Police told the Supreme Court its facial recognition system flagged 2,873 people with criminal records at July's Jantar Mantar exam-leak protests; The Indian Express, citing police, jail and court records, found at least 25 of them, accused of murder, attempted murder or rape, were in jail at the time. Police say any action requires "field verification".

Hong Kong police arrest 15 over AI-forged ID bank accounts

Who did it: HUMAN 東方日報

Hong Kong police arrested two alleged masterminds and 13 nominees who supplied face-verification selfies, saying the group used AI tools to swap their photos onto images of lost or stolen ID cards. Of 210 online account applications, police say 42 succeeded and 14 laundered about HK$1.13 million; the two were charged with two counts of conspiracy to use a false instrument.

Newspapers ask judge to reject OpenAI, Microsoft fair-use defence

Who did it: HUMAN Chicago Tribune

News organisations including the Chicago Tribune and The New York Times asked Manhattan federal judge Sidney Stein to rule that OpenAI's and Microsoft's copying of and training on their reporting was not fair use. In duelling motions, Microsoft called the training transformative and OpenAI argued nobody owns facts; a July sanctions motion accusing OpenAI of destroying evidence is still pending.

Regulators probe Tesla's steering-wheel-free Cybercabs a day after launch

Who did it: HUMAN The Boston Globe

NHTSA opened an investigation into whether Tesla's two-seat Cybercabs, which lack steering wheels, mirrors and brake pedals, comply with federal safety rules, a day after Elon Musk launched rides in them in Austin. The agency will audit the self-certification process Tesla relied on; Tesla stock fell nearly 6 percent on Friday.

Michigan court strikes, denies pro se motions bearing 'earmarks' of AI

Who did it: HUMAN E.D. Michigan

A federal court in Michigan struck a pro se plaintiff's unsigned summary-judgment motion against Anchor Bay Packaging Corporation and denied three others, saying they broke local rules and a pretrial order barring such filings and had "all the earmarks" of AI generation. The plaintiff must show cause by September 25, 2026 for ignoring the order or face possible sanctions or contempt.