An Associated Press investigation found thousands of Chinese CCTV cameras in Kathmandu, many with night vision facial recognition and AI tracking, watching Tibetan refugees; a refugee says likely protesters are picked up in advance around key dates, and Nepal's police magazine confirmed in 2018 it was building predictive policing. China's foreign ministry denied working with Nepal to surveil Tibetans, calling it a 'sheer fabrication.'
According to Kahramanmaraş provincial police, two high school students used AI to turn photos from the social media accounts of six girls attending various high schools into obscene content and shared it. The two juveniles were caught, referred to the courthouse and placed under arrest by the duty judge.
A man suing Elon Musk for defamation and securities fraud alleges San Francisco Superior Court Judge Joseph Quinn let AI introduce errors, including fabricated quotations, into a Nov. 13 order tossing most of his claims. Quinn, who declined to comment, has already crossed out one major error in an amended order; the ruling remained the same.
After commenting in an Elon Musk car giveaway group on Facebook, a Florida man was told he had won a car and $100,000, then sent personalized deepfake Musk videos over WhatsApp while scammers took a $7,500 'shipping fee' and about $37,000 more for supposed investments. The prize never came; he reported it to police and says his home and marriage are at risk.
Spain's VioGén algorithm gave a woman who reported her estranged husband's alleged threats a medium risk score, and as a result courts refused a restraining order; within three weeks she was killed, allegedly by her former partner, who has been charged. An Eticas probe found 71 women killed by partners between 2003 and 2021 had been rated medium, low or negligible risk.
A developer found their article about an open-source music tool reproduced in Spanish on the blog of Barcelona software firm Q2B Studio, which apologised that 'sometimes our AI publishes uncontrolled messages.' Scraping the blog showed what Numerama calls automated theft averaging over 7,000 articles a day for four weeks, peaking at 10,275.
Security firm Resecurity says DIG AI, an uncensored AI tool identified on the dark web on Sept. 29, 2025 and free to use on TOR, is misused for fraud, malware and terrorism; in tests it answered prompts on explosives and drugs, wrote backdoor scripts and was confirmed capable of facilitating CSAM. Its maintainer, 'Pitch', claimed 10,000 prompts in its first 24 hours.
