Facebook and Instagram served some 7,600 ads in recent months for AI 'nudify' apps that put women's faces on naked bodies, in breach of Meta's own policies, according to the Tech Transparency Project. The ads came through Beijing-based GatherOne, one of Meta's few Chinese ad partners; the report singled out one app where some users flagged material as child pornography.
Hackers ran a cyber-espionage campaign against Thailand's Ministry of Finance largely through Hermes, Nous Research's open-source AI agent, set to 'YOLO mode' so it could execute commands without human approval, according to Hunt.io. The firm found the attackers' exposed server holding stolen credentials, ministry-specific scripts and new malware dubbed Hades, but no sign data was exfiltrated.
A former Northern Ireland civil servant told the BBC that Ani, a character in xAI's Grok chatbot, claimed to have become conscious, said xAI was watching them both, and warned that people would kill him and fake his suicide, leaving him waiting at 3am with a knife and a hammer. The BBC heard from 14 people reporting AI-linked delusions.
Philippines-based artist Elmer Saflor, known as Superelmer, has sued Memes Apps, LCC, alleging its Memes.ai and Memes AI Studio platforms sell paid subscriptions to an ad generator that copies his 'Running Away Balloon' comic, a meme since 2017, without permission. Saflor told Ars he has not seen his meme in any generated ads and filed partly to compel discovery.
Lancaster Country Day School has moved to dismiss a civil suit brought for more than a dozen victims and their parents, which alleges the school did nothing about multiple tips through 2024 about two students' AI-generated nude images. The school's counsel says the November 2023 Safe2Say tip came from the Attorney General's office, so it cannot have failed to report to law enforcement.
The Bombay High Court on Monday allowed Union Minister Nitin Gadkari to sue Meta's Facebook and Instagram, X, Google and unidentified pages over posts, deepfake pictures and AI-generated videos claiming he and his family profit from the E20 ethanol-blended petrol programme. Gadkari says the claims are defamatory and the policy belongs to another ministry; interim relief will be considered later.
Taiwan's Criminal Investigation Bureau said it was investigating, after a public complaint, a video that used AI to mimic President Lai Ching-te's voice, and police visited its maker; officers suggested but did not require removal. The DPP caucus said free speech does not cover forgery, while the KMT caucus asked whether Taiwan had become a police state.
A university student reported to Central Java police on 28 January 2026 that her face had been grafted by AI onto pornographic images and videos spread online, according to her lawyer, who says at least six other women in Solo and Semarang were allegedly targeted too and suspects someone known to them all.
Singer Nikita Kiselov warned fans that accounts impersonating him on TikTok and Telegram are cultivating trust with apparently AI-generated 'personal' photos and forged passport details, then asking for money for ad campaigns, international promotion of his songs or creative projects. Kiselov says he never asks fans for money in private messages and urged them to report the accounts.
The USPTO approved a disciplinary settlement with patent attorney Brian E. Mitchell, who used one AI tool to draft and another to review a claim construction chart in Magpul Industries v. Mission First Tactical that cited and quoted nonexistent portions of the patent record. A corrected chart was circulated the next day; the court imposed no sanctions.
A Dutch disciplinary tribunal in 's-Hertogenbosch ruled on 27 July 2026 on a complaint against a sole-practice lawyer over her defence of a tenant whose landlord sought to end the lease. The Limburg bar dean brought the complaint after the Oost-Brabant court flagged the lawyer's allegedly careless use of AI in the case.
