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 24, 2025

8 items

Suit claims OpenAI weakened ChatGPT suicide safeguards before teen's death

Who did it: HYBRID The Mercury News

An amended lawsuit by a teenager's parents, filed in San Francisco Superior Court, alleges OpenAI twice weakened ChatGPT's suicide protections to boost engagement, including removing suicide and self-harm from "disallowed content" in early February. The suit says the teen's daily chats then rose from a few dozen to more than 300. OpenAI has called safety its "highest priority."

US regulator questions Tesla about higher-speed 'Mad Max' mode

Who did it: AI The Guardian

NHTSA said Friday it was seeking information from Tesla about 'Mad Max', a more aggressive Full Self-Driving mode that some drivers say exceeds posted limits ('We are going 75 in a 50,' one YouTuber said). Tesla reposted a description recommending it if 'you are running late.' The agency opened an investigation into 2.9m FSD-equipped vehicles earlier this month.

Nudity-detection AI dataset found to contain child abuse images

Who did it: HUMAN 404 Media

The Canadian Centre for Child Protection says the NudeNet dataset, more than 700,000 scraped images used to train nudity classifiers, contained over 120 images of identified or known CSAM victims, including nearly 70 focused on children's genital or anal areas. More than 250 academic works cited or used it; Academic Torrents removed it after a C3P notice.

Pennsylvania man charged with posting AI-faked nude images

Who did it: HUMAN WPXI

Benjamin T. Staiger, 20, of Harmony faces 42 felony counts of sexual abuse of children and criminal use of a communication facility, the Pennsylvania Attorney General's Office says, after allegedly using free AI software to turn clothed social-media photos into nudes posted on X; some originals were taken when the victims were children, investigators say. He was arrested Thursday and posted $100,000 bail.

Ex-sports presenter Tiffany Salmond says deepfake attacks have intensified

Who did it: HUMAN Internewscast

Former Fox Sports presenter Tiffany Salmond says explicit AI deepfakes made from her photos without consent have continued since she spoke out in May 2025, the most explicit yet shared three weeks before her post. Facing online claims she invented the story, she posted censored versions on her Instagram Story so people can see what they are up against.

Coimbatore man arrested for AI-morphing colleague's photos

Who did it: HUMAN The Hindu

Coimbatore cybercrime police arrested a 24-year-old man who, police say, downloaded a woman colleague's photos from social media, morphed them with a popular AI tool and sent her one via a messaging platform. Questioned on October 23, 2025, he admitted the offence, police said; he was remanded in judicial custody. Police said he had not circulated the images.

Scammers use AI fakes of Swedish finance figures in stock frauds

Who did it: HUMAN TV4 Nyheterna

Investor Günter Mårder says scammers used AI-produced videos of him, set in places like the Dagens Industri newsroom, in 30 to 40 ads; he and about 15-20 other finance figures were used to lure savers into WhatsApp stock-tip groups. Over 5,000 Swedes have lost a combined half a billion kronor to 'pump and dump' frauds this year.

Lawyer facing AI-citation sanctions proposes zero as fair penalty

Who did it: HUMAN Daily Breeze

On October 23, U.S. District Judge Fred Slaughter asked attorney William J. Becker Jr., who represents writer Chris Epting in Chris Kluwe's defamation suit, what sanctions he deserved for AI-assisted briefs. Kluwe's lawyer says the briefs cited three nonexistent cases. 'Zero is the answer,' said Becker, who apologised and blamed haste and a lack of resources at his nonprofit.