A 42-year-old serviceman based in Moselle was charged in Metz and remanded in custody, accused of making "more than 10,000 requests" to force the Ministry of the Armed Forces' AI tool to generate child sexual abuse images, deputy prosecutor Thomas Bernard said. Two colleagues have also complained of nude images of them made with the tool; the suspect, arrested 22 July, admitted the facts.
Acting on a complaint, Italy's data protection authority found R.T.I. Spa breached GDPR articles 5 and 25 with Striscia la Notizia segments using AI to manipulate a well-known journalist's image and voice into statements never made. Noting that even satirical deepfakes must respect dignity, it ruled the disclaimers unclear, issued a formal warning and barred further use of the journalist's data that way.
The Dutch Advertising Code Committee ruled that Philip Morris breached the tobacco advertising ban with its "your voice your choice" campaign, in which QR codes in tobacco shops led to a company-owned AI tool that wrote first-person objections to stricter EU tobacco rules. Earlier research by NOS and Pointer found nearly three-quarters of Dutch consultation responses were fully AI-written with that tool.
Homo Digitalis, the Institute of Regulatory Policies and a civil servant asked Greece's National Transparency Authority to investigate whether the Interior Ministry lawfully used an AI tool to reprocess 2023 evaluations of 54,500 civil servants rated high-performing. The complaints say fewer than 4,500 kept that rating and 1,329 were judged low-performing, and allege no specific legal basis existed or impact assessment was published.
According to the Hong Kong police's Cyberdefender platform, cited by the South China Morning Post, scammers used AI-made WhatsApp voice messages matching a man's father's voice and manner to request urgent HK$1 million transfers. The victim kept paying until HK$10 million, the entirety of the savings, was gone; police now advise: "Do not blindly trust voice messages."
After a Hinatazaka46 member released a first solo photobook, apparently AI-altered versions of its swimwear and lingerie shots, some exaggerating body features, circulated online alongside unauthorised reposts and sales. Publisher Shueisha said on 6 August that all the photos are copyrighted, that unauthorised editing or distribution is illegal, and that it may seek takedowns and legal action if infringement continues.
Spanish fact-checker Maldita.es identified more than a thousand YouTube channels where AI-generated doctors spread health disinformation in over 30 languages, 20 of them in Polish, claiming for example that eggs cure cancer and diabetes and that sunscreen causes cancer. Poland's NASK says accounts it reported were removed, but it received no response regarding the YouTube accounts.
Arbitrator Ryan P. Newell denied Counsel Stack's motion for leave to seek dismissal of LawCatch's claims over a $375,000 citation-verification system meant to use an LLM to detect misinterpreted case law. LawCatch, which seeks its money back, had argued that the authority underlying the motion did not stand for the stated proposition.
