Meta ran 32 ads, found by the Tech Transparency Project, for Kromix, an AI app promising 'no restrictions', including one in which a woman resembling a prominent female US politician winks before a cut to pornographic footage, despite Meta's ban on sexual ad content. The ads came down after WIRED asked, and Apple said it had removed the app from its App Store.
Prosecutors in Gießen are investigating a 32-year-old daycare worker suspected of using AI to turn photos of children into realistic child sexual abuse images and videos, after a tip from the US National Center for Missing & Exploited Children. His employer fired him; prosecutors say there is no solid evidence of physical abuse, and he remains free.
Brazil's Justice Ministry asked the Federal Police and data protection agency ANPD to assess blocking 80 websites offering AI tools that generate fake nude images and videos of women, citing signs of serious violations of the rights of women, children and adolescents. The request followed a filing from SaferNet Brasil listing the sites.
Paris prosecutors opened an investigation into a suspected Russian disinformation campaign using deepfakes and fake video reports on X against former prime ministers Gabriel Attal and Edouard Philippe, centrist candidates for 2027. Posts impersonating French media falsely claimed Attal had Parkinson's disease; watchdog Viginum attributed the Attal campaign to the pro-Russian Matryoshka network with high confidence.
After BLACKPINK's Jennie closed Summer Sonic 2026 in Chiba, a clip appearing to show a wardrobe malfunction spread online, prompting debate over whether it was real or maliciously AI-faked. Her agency, OA Entertainment, says people have maliciously edited her photos and videos, and that it is gathering evidence for civil and criminal action, with no settlement or leniency.
A federal judge in Tacoma excluded two of a plaintiff's experts and dismissed her remaining claims against supplement maker Outliers, having earlier sanctioned her attorney, Jocelyn Stewart, for submitting unverified AI-generated false citations. The plaintiff's side admitted using Claude or ChatGPT for a citation table appended to two expert reports; defendants said a newly excluded expert's report had similar errors.
Lawyers for Imperial Reliance in a Dallas federal case sought leave to fix a July 29 brief, partly prepared with generative AI, after opposing counsel flagged a miscited case name and quotations absent from the cited decisions. Counsel said a premium AI Westlaw subscription had been used to verify the citations, and that a new manual process is now in place.
A restaurant owner in Galicia says a couple tried to avoid paying a bill of hundreds of euros by presenting a suspicious video, according to a Greek news report, which says restaurateurs complain that customers use allegedly manipulated or possibly AI-generated content to dispute their bills. The report gives no further detail about the video.
