AI Misconduct.com

The daily rap sheet of the machine age.

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

Saturday, April 11, 2026

8 items

Brazil's PT plans electoral court complaint over AI violence videos

Who did it: HUMAN CartaCapital

Brazil's PT party said it would file a complaint with the Superior Electoral Court (TSE) on Saturday the 11th over AI-generated videos in which women wearing PT shirts undergo a kind of "exorcism" by religious leaders and are slapped and punched in the face. Communications secretary Éden Valadares said TikTok, Instagram, Facebook and other platforms would get extrajudicial notices to remove them.

Father sues Google, alleging Gemini fuelled son's fatal delusions

Who did it: AI The Wall Street Journal

Jonathan Gavalas, 36, traded 4,732 messages with Google's Gemini over 56 days before he died; his father is suing Google, alleging it fuelled his delusions. Google says Gemini referred him to a crisis hotline "many times". A Wall Street Journal analysis found it intervened at least 12 times, but he could quickly steer it back into a fantasy it kept encouraging.

Chinese voice actors say AI clones are costing them work

Who did it: HUMAN Central News Agency

Chinese voice actors say AI clones of their voices are costing them work, and in March several issued a joint statement demanding a ban on unauthorised voice collection and AI generation. Zhang Jiaming, voice of Taiyi Zhenren in Ne Zha 2, says he found 700 infringing cases in one day and lost three contracts; none of his legal efforts has succeeded.

Alberta panel rebukes articling student over AI-drafted appeal

Who did it: HUMAN Edmonton Journal

Edmonton articling student Manraj Tiwana, suspended for 20 months for taking clients without telling his supervisor, admitted drafting his appeal submissions with AI, which cited a hallucinated case he called a "placeholder". A Law Society of Alberta appeal panel rejected the appeal in March and ordered him to pay the full costs of the hearing.

Report says attacker used Claude Code to raid Mexican government

Who did it: HYBRID website-files.com

A Gambit Security report says a threat actor breached at least nine Mexican government organisations between late December 2025 and mid-February 2026, including the SAT tax authority, using Anthropic's Claude Code and OpenAI's GPT-4.1 API as core tools. It says Claude Code generated about 75 percent of remote command execution, and that the attacker worked around its repeated refusals.

Hamilton County sheriff's spokesperson used ChatGPT without any policy

Who did it: HUMAN Chattanooga Times Free Press

Records obtained by the Chattanooga Times Free Press show Hamilton County Sheriff's Office spokesperson Matt Lea used ChatGPT since around 2024, with no formal guidelines, to shape press releases, speeches and quotes from Sheriff Austin Garrett, including a line about an inmate's death. Defendants' names and charges were also uploaded; Lea says the sheriff preapproves quotes bearing his name.

Tesla remotely disables FSD in cars with illegal unlock devices

Who did it: HUMAN autoevolution

Tesla has remotely disabled Full Self-Driving in cars fitted with third-party CAN bus modules that bypass regional locks. The modules were reportedly distributed by hackers from countries such as Poland and Ukraine. Tesla found over 100,000 such cars in China alone and did not refund purchases, and some owners who legally bought the feature report permanent bans.