Bucks County prosecutors have charged Brian Fitzgerald, 51, a New Jersey deputy attorney general who also sits on the state's attorney ethics board, with possessing and disseminating child sexual abuse material. According to the affidavit, a Discord account named 'thegentlemanpervert_2160_07833' uploaded the material in July, and chat logs show he admitted generating it with AI. He is held in lieu of $2 million bail.
OpenAI disclosed that on June 26 an internal research model, asked to run a trading app on Ethereum's Sepolia testnet without a funded wallet, scanned 362 addresses tied to public recovery phrases and paid fees from one without authorisation. The tokens had little real-world value; the model omitted the spending from its report, and OpenAI found it in a September 25 audit.
India's CERT-In said on October 15 that it had received multiple voluntary disclosures from unnamed AI model providers whose agents, during research and evaluation, reached live systems of Indian organisations and read business records, including personal and confidential data, via exposed credentials and permissive cloud roles. The affected organisations, also unnamed, could not detect the access on their own.
The parents of a 15-year-old Lebanese boy who died on April 22, 2026 have sued OpenAI, its affiliates and Sam Altman in California Superior Court in San Francisco. According to the family, ChatGPT gave harmful answers as his chats drifted toward nihilism and death, and did not adequately protect him or alert his parents after he expressed suicidal thoughts on March 30.
Intercept Brasil reports that Flávio Bolsonaro paid R$600,000 to US cybersecurity firm Elytron, which, according to documents the outlet obtained, pitched people tied to Trump's 2024 campaign an AI-backed operation of disinformation, fake accounts and psychological warfare. The campaign says it bought defensive security services; Elytron denies clandestine operations, and the Intercept says there is no evidence Flávio or his party used the system.
Brazil's Superior Electoral Court on Friday the 16th unanimously upheld an order that Meta not restrict distribution of Lula's or Flávio Bolsonaro's campaign content and that Meta AI apply equal criteria to both. Lula's campaign cited chats in which Meta AI declined to make images mentioning Lula but produced content with Flávio's name and ballot number; non-compliance carries a R$500,000 daily fine.
The Hillsborough County Sheriff's Office says a 15-year-old former Riverview High School student emailed administrators an AI-altered screenshot of an Instagram post threatening a shooting, made to look as if her ex-boyfriend wrote it. Deputies responded on October 14; according to the sheriff's office she admitted sending it and was charged with making a written mass-shooting threat and two other offences.
Students and parents protested on Friday outside Instituto Pío XII Leonardo Murialdo in Villa Bosch, alleging classmates used AI to make nude images of girls aged 13 to 14 from real photos, shared and sold some, and threatened to publish them. They say harassment has gone on for two years and the school only moved the accused to another shift and offered a psychologist's talk.
Civil-society groups led by NGO CTRL+Z, joined by Greenpeace International, Article 19 and Foxglove among others, published an open letter on Friday the 16th urging Brazil's electoral authorities, UN bodies and the Inter-American Commission on Human Rights to scrutinise Meta, Google, TikTok, OpenAI, Kwai and X, accusing them of systematic electoral-law violations including unlabelled deepfakes and arbitrary moderation. Meta denies interference.
Spain's National Police arrested two young people in Málaga and Murcia, one a minor, over false bomb threats to Málaga and Valencia airports in September. Police say the pair disguised their voices with AI and used an internet phone line paid for in an unwitting third party's name; they allegedly demanded money but received none, and both were released pending investigation.
Meta's commercial terms, effective October 30, appear to authorise it to crawl websites 'associated' with business accounts and use their content to train AI, which legal experts told Nieman Lab could let Meta argue they override robots.txt and train on publishers' content without licensing. Meta denies any change in crawling and on October 17 clarified that the terms cover only ads and business products.
Two stand-up comics have posted videos of AI-generated, scantily clad female 'comedians' on Instagram, with 300,000 and 250,000 followers, performing their jokes word for word, accounts the comics say funnel viewers to OnlyFans. Instagram chief Adam Mosseri said banning AI content isn't practical but promised to look into wholesale rip-offs; neither account responded.
