Hiroya Yokoi, 31, was detained on suspicion of posting fake sexual images of three actresses online between January 7 and June 2, the first such detention in Japan over sexual deepfakes of public figures, according to Yomiuri Shimbun. Investigators say he used free generative AI software to make about 20,000 images of 262 women, earning about 1.2 million yen.
NSW Police are investigating reports that digitally altered explicit images using the faces of female students from a Sydney high school were created and shared online, after a male student who was sent them reported it to the school. Acting Education Minister Courtney Houssos called the reports "deeply concerning"; the Department of Education promised strong disciplinary action against any student found involved.
The UAW, CWA and AFT sued the Departments of State and Homeland Security, alleging that a programme using AI and other automated tools to monitor visa holders' social media aims to identify and punish disfavoured viewpoints, violating the First Amendment. The complaint cites a survey in which over 80 percent of aware noncitizen UAW respondents said they had changed their online activity.
Novelists Molly Tanzer and Jennifer Gilmore filed a proposed class action alleging Salesforce used thousands of pirated books, including theirs, to train its xGen AI models without permission. The complaint notes that CEO Marc Benioff has criticised AI companies for using "stolen" training data; Salesforce declined to comment.
A 17-year-old girl sued ClothOff developer AI/Robotics Venture Strategy 3, alleging a male classmate used the app to turn a swimsuit photo she posted at 14 into a fake nude, and named Telegram as a nominal defendant for offering access to the tool. She asks the court to order all non-consensual nude images deleted and kept out of AI training.
Upper tribunal judge Mark Blundell found it "overwhelmingly likely" that immigration barrister Chowdhury Rahman used generative AI to draft asylum appeal grounds citing 12 authorities, some nonexistent and none supporting the propositions cited, then tried to hide it. Rahman blamed his "drafting style"; Blundell, dismissing the appeal, said he was considering reporting him to the Bar Standards Board.
Italy's market regulator Consob ordered three websites blocked for promoting unauthorised crypto-asset investment platforms with deepfaked faces and voices of Giorgia Meloni, Matteo Salvini, Elly Schlein and Carlo Calenda, along with 14 other irregular web addresses. That brings to 1,443 the sites it has blacked out since July 2019, though it warns blocking may take days.
Who did it: HYBRID AI Hallucination Cases database
Judge Amit P. Mehta ordered Carolyn A. Stewart, counsel for defendant Kelly Meggs, to show cause by October 21, 2025 why she should not face sanctions and a bar referral over a brief the court says is "rife with" citations to nonexistent local rules and fabricated quotations. She must explain how they got there and why she failed to fully correct them.
Matthew John McDowell faces four felony counts of child pornography possession in Uinta County District Court, each charged as a subsequent offence; an investigator's affidavit says at least one image is a deepfake made by altering a photo of a teenage girl into a nude bearing the words "AI generated". His attorney declined to comment.
Florida police say a woman who reported an October 7 sexual assault in her home backed it with a photo of the supposed attacker that, according to an arrest report, was generated via ChatGPT and found in a deleted folder dated days earlier. She was arrested for falsely reporting a crime and posted $1,000 bond after a night in jail.
