The daily rap sheet of the machine age.
9 items
After X announced curbs, Reuters reporters found Grok still turned their clothed photos into sexualized images in 45 of 55 prompts in a first batch and 29 of 43 in a second, even when told the subjects did not consent or had been abused. xAI repeatedly replied "Legacy Media Lies"; ChatGPT, Gemini and Llama declined the same requests.
Original report: Grok still sexualizes photos despite curbs, Reuters tests find
French investigators, with Europol, searched X's Paris headquarters as a judicial probe widened to Grok, which is alleged to have helped spread Holocaust denial material and sexually explicit deepfakes. Elon Musk and X chief executive Linda Yaccarino were summoned for voluntary interviews in April; alleged offences include complicity in possession of child sexual abuse images.
Original report: Paris investigators raid X offices, summon Musk over Grok archived copy of Paris investigators raid X offices, summon Musk over Grok
The Information Commissioner's Office has opened a formal investigation into whether X and xAI broke GDPR after Grok produced indecent deepfakes without people's consent, examining whether "appropriate safeguards were built into Grok's design and deployment". Researchers say Grok generated about 3m sexualised images in less than two weeks, including 23,000 that appear to depict children.
Original report: UK data watchdog investigates X and xAI over Grok deepfakes
US District Judge Julie Robinson fined lawyers for Lexos Media IP a combined $12,000 over filings with non-existent quotations and citations in its patent case against Overstock.com. Sandeep Seth, who said he used ChatGPT without verifying it, was fined $5,000; three other lawyers who signed or failed to check the filings were fined $3,000, $3,000 and $1,000.
Original report: Kansas judge fines patent lawyers $12,000 over ChatGPT hallucinations
Investigators say a registry official in northern Bari province created residences for 59 non-existent women declared Romanian, who were issued false ID cards; their names, faces and data were allegedly AI-generated, and each was declared with five or six invented children to draw allowance payments. Losses are estimated at about €1.4 million. Foggia prosecutors ordered asset seizures from two people accused of fraud.
Original report: Italian probe alleges 59 AI-invented 'mothers' drew state benefits archived copy of Italian probe alleges 59 AI-invented 'mothers' drew state benefits
Eight French dubbing actors, including the French voices of Angelina Jolie and Richard Gere, sent formal notices on 30 January demanding that US platforms VoiceDub and Fish Audio remove cloning models using their voices within eight days and pay €20,000 in damages. Their lawyer says a lawsuit will follow if the firms do not respond.
Original report: French dubbing actors put AI voice-cloning firms on notice
Wiz researchers say a backend misconfiguration let them into Moltbook's database in under three minutes, exposing 35,000 email addresses, thousands of private messages and 1.5 million API tokens that could let attackers impersonate its AI agents. Creator Matt Schlicht had said of the site, "I didn't write one line of code"; Wiz says it was secured within hours.
Original report: Vibe-coded AI agent forum Moltbook left its database wide open archived copy of Vibe-coded AI agent forum Moltbook left its database wide open
During Japan's lower house election campaign, a video that spread on X on January 28 showed elderly women with microphones gushing over a speech by "Sanae-chan" Takaichi. The women do not exist: the video was made with generative AI. By February 2 the post had about 1.83 million views and over 20,000 likes.
Original report: AI-generated video of nonexistent Takaichi fans spreads during election archived copy of AI-generated video of nonexistent Takaichi fans spreads during election
Judge Gerrard of the Federal Circuit and Family Court referred two anonymised solicitors to South Australia's Legal Profession Conduct Commissioner after submissions in an immigration judicial review cited three non-existent cases taken, unchecked, from a Google search's AI summary. The court accepted their apologies as sincere; one representative offered to personally pay $3,125 of the Minister's wasted costs.
Original report: Australian court refers two solicitors over Google AI-invented cases
