Magistrate Judge Thomas Hixon ordered Meta to identify executives who authorized torrenting books from shadow libraries like LibGen and Anna's Archive for AI training, in suits by authors and Entrepreneur Media. Meta argued its early data gathering was an 'informal approach' and a 'decision' was 'made only in the sense' that staff 'technically' downloaded torrents; one staffer testified to needing director or VP approval.
TikTok staff warned leadership about minors stripping on Live for 'gifts' convertible to cash and about algorithmic 'rabbit holes' with a 'real contagion effect' on suicidal ideation, according to depositions filed in federal court in March and reported by NPR. One safety lead said they had been 'fighting about fixing this for months'; the filings belong to multi-district litigation alleging platforms concealed harms to children.
Login credentials for Smart Sampa, São Paulo's AI-assisted camera network, sat in a public online repository for almost a year, according to UOL, exposing more than 1,000 cameras at health clinics and 427 at preschools. Mayor Ricardo Nunes said the system lacked two-factor authentication, blamed a suspended municipal guard, denied improper access and said the city knew of the exposure, which served as 'bait'.
Meta paused sharing in its Muse AI assistant after security firm Claroc found more than 6,800 shared-chat links through public DNS records. Because each address was built from the conversation's title plus a random ID, Claroc says, a hostname from DNS hinted at the contents and could rebuild the link; exposed chats included a loan application, a company revenue review and a staff calendar.
Meta's Oversight Board said content moderation failures led to the removal of 'No Kings' protest accounts in the US and accounts focused on Brazil's election, adding that No Kings accounts should have been in Meta's cross-check programme. Meta called the removal an automated action 'clearly taken in error'; separately, Brazil's top electoral court doubled a daily fine for non-compliance with an order on political content.
The 21-year-old accused of stabbing a 19-year-old woman more than 40 times in Ciudad del Plata on 3 June, who investigators say used AI to research how to kill and the likely sentence, will be declared not criminally responsible after forensic assessments. A simplified trial will order rehabilitation at Hospital Vilardebó; the family's lawyer disputes the findings.
BBC News Brasil mapped 37 Instagram and TikTok profiles posting AI-generated women backing Flávio Bolsonaro or Lula, together drawing about 7 million followers; some steer fans to OnlyFans or Telegram to sell pornography. Most lacked the AI labels the platforms require, and one persona was edited into a scene with Flávio in place of a real congressman. TikTok said it removed the reported content.
Purported intimate photos of a TXT member spread on Weibo on 19 October; fan groups argue they are AI fakes, citing the wrong hair colour for the US tour, a missing arm mole, a Chinese-standard power socket in a supposedly American setting and a third-party detector's 79% generation score. BIGHIT MUSIC had not commented by that evening.
A South Carolina mother says her 6-year-old daughter came home from elementary school with an AI-generated image of herself at 25, dressed as a nurse. In a viral post, the mother raised privacy concerns; she asks what software was used, what happens to the photos and whether anyone consented, and says she is still awaiting answers from the principal and school board.
