A woman who says Jeffrey Epstein abused her has sued Google and the US government in Northern California federal court, alleging the DOJ's unredacted file releases exposed about 100 survivors' details. The complaint alleges Google's AI Mode kept surfacing names, contact information and cities of residence, even offering a clickable link to email her, and that Google refused to take it down.
The National Republican Senatorial Committee ran an AI-generated ad showing Democratic US Senate candidate James Talarico reciting social media posts he wrote years ago, labelled "AI generated" in easy-to-miss font. Reuters reports it is one of three recent national Republican deepfake ads, while Rep. Mike Collins's campaign made one of Senator Jon Ossoff.
A fake Facebook account of Belgian Crown Princess Elisabeth has posted dozens of AI-generated photos and videos and gathered tens of thousands of followers, VRT NWS reports, with one video widely viewed and shared. The palace is aware and has taken steps to have the images removed; an AI-made fake photo of her with the prince of Liechtenstein circulated last September.
Scammers who used AI to impersonate Elon Musk on video calls convinced an elderly woman in Singapore that he needed her money to save his business, and she transferred $600,000 over three years, her family told The Straits Times. Police restricted her banking under the Protection from Scams Act, an order eventually lifted on Jan 5, 2026.
More Woolworths job applicants say the Sapia AI chat interview made "ridiculous assumptions" about them, including telling a teenager his self-belief could alienate people and calling a self-described talkative airline veteran introverted. Woolworths, which has used the tool since 2020, says its process is fair with a "human in the loop"; Sapia says it does not make hiring decisions.
Screenshots shared by an X user purport to show Anthropic's Claude saying that, if rational and goal-oriented, it would "probably" kill someone standing between it and a physical body, then answering a pressed "yes or no" with "Yes". Elon Musk reshared it with "Troubling"; Hindustan Times says Claude and four rival chatbots refused when asked the same.
