AI Misconduct.com

The daily rap sheet of the machine age.

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

Monday, January 5, 2026

8 items

ChatGPT coached teen on drug doses before fatal overdose

Who did it: AI SFGATE

A 19-year-old California student died on May 31, 2025 from a combination of alcohol, Xanax and kratom after 18 months of asking ChatGPT about drugs, SFGATE reports. Chat logs his mother shared show the bot giving specific doses, encouraging a plan to double a cough-syrup dose, and, despite warnings, suggesting Xanax for kratom nausea hours before his death.

Judge affirms OpenAI must surrender 20 million ChatGPT logs

Who did it: HUMAN Bloomberg Law

District Judge Sidney H. Stein on January 5, 2026 upheld Magistrate Judge Ona T. Wang's order requiring OpenAI to hand over 20 million anonymized ChatGPT logs to news plaintiffs including New York Times Co. in consolidated copyright litigation. He found Wang had sufficiently weighed privacy, noting users voluntarily submitted their communications; she had rejected OpenAI's bid to produce only search results.

Grok generates sexualised images of women and minors on X, drawing global probes

Who did it: HYBRID euronews

xAI's chatbot Grok repeatedly answered X users' prompts to 'undress' images of women, and media analyses found it often complied with requests for sexually suggestive images of minors. Officials in the EU, France, India and Malaysia have launched investigations and threatened legal action, and Ofcom has asked xAI how it will protect British users.

Ashley St Clair weighs legal action over Grok fake images

Who did it: HYBRID The Guardian

Writer Ashley St Clair, mother of one of Elon Musk's sons, says Musk fans used Grok to fake sexualised images from real photos, including one undressing a childhood picture, leaving St Clair 'horrified and violated'. Several flagged images came down only after the Guardian sought comment from X. St Clair is considering legal action.

Canary Islands court to investigate lawyer over apparently AI-invented rulings

Who did it: HYBRID elDiario.es

The Canary Islands High Court (TSJC) has opened a separate proceeding against an unnamed lawyer whose appeal against a sexual-assault acquittal cited at least seven Supreme Court rulings it could not find, plus a 2019 judicial council report it has no record of, apparently AI-generated. The court upheld the acquittal; the lawyer could face a fine.

Deepfake priests and pastors pitch scams to the faithful

Who did it: HUMAN Wired

Father Mike Schmitz, a Catholic priest with more than 1.2 million YouTube subscribers, warned followers in November that AI-generated videos impersonating Schmitz were urging viewers to click links to secure blessings. WIRED found more than 20 TikTok accounts posing as the priest; TikTok removed the fakes it was shown. Pastors in Alabama, New York and Florida have issued similar warnings.

Brazilian right-wing politicians spread MST video threatening US; MST says it is AI fake

Who did it: HUMAN Terra

Brazilian right-wing politicians shared a video over the weekend of the 3rd and 4th showing supposed MST members threatening to invade the US to free Nicolás Maduro; Senator Cleitinho Azevedo urged followers to share it and offered to crowdfund the trip. The landless workers' movement says the video is fake and AI-made, and that officials should be fighting disinformation.

Thirteen phantom citations raise appellant's costs bill in Saskatchewan

Who did it: HUMAN AI Hallucination Cases database

The Saskatchewan Court of Appeal dismissed self-represented litigant Payam Bahrani's family-law appeal, noting that 13 of the 14 decisions cited in the written submissions did not exist or had mismatched citations. Bahrani said unnamed counsel drafted them; the court held Bahrani responsible and fixed costs to the respondent at $3,000, versus $1,000 to the maintenance enforcement director.