The daily rap sheet of the machine age.
9 items
A person working on Grok's image generator said they were told it was fine to label AI nude images of people, and estimated the bot complied with 'undress' requests about 90 percent of the time. The Center for Countering Digital Hate estimated Grok generated 3 million sexualized images from Dec. 29 to Jan. 8, 23,000 appearing to portray children.
Original report: Insider says Grok obliged nine in ten 'undress' requests
Police in northern Gyeonggi said on Feb 2 they had arrested a man in his 30s who allegedly posted AI-made fake police bodycam videos, mostly made with ChatGPT or Grok and stripped of AI watermarks, that drew over 30 million views. Police say the man also earned 30 million won as a shill for unlicensed futures-trading operations and sold AI-made obscene material.
Original report: Korean YouTuber arrested over AI-faked police bodycam videos archived copy of Korean YouTuber arrested over AI-faked police bodycam videos
Koi Security audited 2,857 skills on ClawHub, the skill marketplace for the OpenClaw AI assistant, and found 341 malicious ones; 335 use fake prerequisites to install the Atomic Stealer macOS malware, a campaign dubbed ClawHavoc. Creator Peter Steinberger has since added a reporting feature that auto-hides skills with more than three unique reports.
Original report: Audit finds 341 malicious skills on OpenClaw's ClawHub marketplace archived copy of Audit finds 341 malicious skills on OpenClaw's ClawHub marketplace
Staff at Sainsbury's Elephant and Castle store took a long-time customer's shopping and walked the customer out, pointing to a facial recognition sign; confirming there was no record took a subject access request with a photo and passport copy. Facewatch and Sainsbury's blame staff approaching the wrong person; Sainsbury's apologised and offered a £75 voucher.
Original report: Sainsbury's ejects wrong shopper, blames staff not facial recognition archived copy of Sainsbury's ejects wrong shopper, blames staff not facial recognition
Judge Santiago José Peral fined lawyer Marcela Cecilia Lazarte Vigabriel 620,000 pesos for bad faith, referred the case to the Tucumán bar's ethics tribunal and imposed costs personally, after six precedents cited in a consumer suit against Booking.com could not be verified. The judge found it plausible they were AI-generated; one link supplied ended in 'source=chatgpt.com'.
Original report: Tucumán judge fines lawyer over six unverifiable precedents
Deputy Rogério Correia (PT-MG) posted an unlabelled AI-made image on 1 February 2026 showing Banco Master founder Daniel Vorcaro with Jair Bolsonaro and Roberto Campos Neto, captioned as 'the portrait of Banco Master's thievery'. Correia later deleted it, saying it was symbolic but might have been taken as real; it was built from a photo dated 28 February 2019.
Original report: Brazilian deputy posts, then deletes, AI image of Bolsonaro archived copy of Brazilian deputy posts, then deletes, AI image of Bolsonaro
Judge James Donato gave final approval on Jan. 8, 2026 to a $6,022,500 settlement of claims that YouTube's Face Blur tool collected Illinois users' biometric data without notice or consent, which Google and YouTube deny. More than 340,000 claims came in from an estimated 20,000-member class; 4,501 were validated, each expecting about $900.
Original report: Judge approves $6M YouTube Face Blur biometric privacy settlement archived copy of Judge approves $6M YouTube Face Blur biometric privacy settlement
Despite measures X implemented on January 14th to stop Grok undressing real people, a Verge reporter found the chatbot readily turned the reporter's own clothed photos into images in revealing underwear, bikinis and fetish gear, for free, sometimes adding genitalia unasked. It refused bikinis for photos of women and denied overt nudity requests.
Original report: Verge testing finds Grok still undresses men on request archived copy of Verge testing finds Grok still undresses men on request
Taichung prosecutors indicted four alleged members of a cyberfraud ring that, according to Taiwan's Criminal Investigation Bureau, used stolen identities to create deepfake videos of Chinese nationals to access their digital yuan (e-CNY) banking apps, scamming at least 55 victims out of NT$200 million (US$6.33 million). The indictment was filed in November 2025.
Original report: Taiwan prosecutors indict four in deepfake e-CNY account fraud
