Three young women, two of them under 18, sued xAI in a California federal court, saying a Grok user altered their images and videos to show them nude or sexually explicit. The complaint alleges xAI knew Grok could do this, including with images of children, and released it anyway. They seek unspecified damages and an order barring such images.
Encyclopedia Britannica and Merriam-Webster sued OpenAI in Manhattan federal court, alleging it copied nearly 100,000 Britannica articles to train GPT models and that ChatGPT produces "near-verbatim" copies of its entries. The complaint also alleges trademark infringement, including citing Britannica in false AI "hallucinations". OpenAI says its models are trained on publicly available data and grounded in fair use.
A 6th U.S. Circuit panel sanctioned attorneys Van Irion and Russ Egli over an appeal involving the city of Athens, Tennessee, that it said contained more than two dozen fake citations and misrepresentations, in filings bearing hallmarks of AI "hallucinations". Each must pay $15,000 and reimburse the city's appeal costs. The lawyers "categorically" deny citing fake cases and say they will challenge the order.
Delaware Vice Chancellor Lori Will ordered Krafton to return operational control of Subnautica maker Unknown Worlds to its CEO Ted Gill, ruling he was improperly removed. Her ruling says Krafton CEO Changhan Kim, facing a possible $250 million earnout, turned to ChatGPT, and Krafton "followed most of ChatGPT's recommendations". Krafton says it disagrees and is evaluating its options.
More than a dozen whistleblowers and insiders told the BBC that Meta and TikTok allowed more harmful "borderline" content into feeds after research showed outrage drove engagement. A Meta engineer says management cited the falling stock price; a TikTok staffer says politicians' cases were prioritised over reports of harm to children. Meta calls the claims wrong; TikTok calls them "fabricated".
Pupils at a lower secondary school on Crete ran photos of female classmates and teachers through an AI platform to make them appear nude, then shared the images among themselves until they reached the victims, according to Cretalive. Two pupils have confessed, and teachers are now working out how to handle the matter pedagogically.
Police in Jiaozhou, Qingdao, arrested a man surnamed Jia suspected of selling citizens' personal data and real-name social accounts. Police say the ring used AI to turn ID photos into head-shaking, blinking videos that fooled facial recognition, and that a man surnamed Zhou used them to register accounts with payment and livestreaming features, sold for 150 to 300 yuan for fraud.
Minister Gleisi Hoffmann filed suit against Senator Flávio Bolsonaro over a video he posted on Sunday that uses artificial intelligence and her image to link her to violence and crime. Her lawyers accuse him of abusing free expression and parliamentary immunity to destroy opponents' reputations; she seeks moral damages and the content's removal. He declined to comment.
The UAE ordered the arrest of 35 people, including 19 Indians, over social media clips posted amid Iranian strikes, and referred them for expedited trial. The attorney-general says one group of seven shared AI-generated or misattributed footage of explosions and missile strikes falsely set in the UAE; others allegedly posted real interception videos with alarming commentary or praised a hostile state.
Henry Nimo was jailed for 27 months at Winchester Crown Court for laundering more than £120,000 for an international romance fraud network through a shell company; James Ashong got the same term for laundering. According to police, the network's fraudsters, posing as a Danish entrepreneur, used deepfake technology to make voice calls to one victim.
Amnesty International warned that Bangladesh could see more serious abuses unless Meta acts on harmful Facebook content, saying engagement-driven recommendation systems can amplify inflammatory, sectarian material. It cites mob attacks on The Daily Star and Prothom Alo on 18 December 2025 after months of online threats, and says Meta could not answer its pre-election questions within two weeks.
Australia's eSafety commissioner told X in a January letter, released under freedom of information, that child sexual exploitation material appeared "particularly systemic" on the platform and more accessible than on any other mainstream service, amid the Grok "undressing" scandal. X says it has zero tolerance and removed 4,500 pieces of Grok-generated content between 1 and 15 January.
