An Electronic Frontier Foundation analysis of Flock Safety search logs found more than 50 federal, state and local agencies ran hundreds of licence-plate reader searches linked to protests between December 2024 and October 2025, often citing just "protest" as the reason. Three agencies, including Delaware State Police and the California Highway Patrol, also targeted animal-rights group Direct Action Everywhere, the EFF says.
A cheating ring used computer-synthesised photos blending candidates' and stand-ins' faces for fake IDs, letting stand-ins sit national exams, including the civil service exam, 89 times between 2014 and 2020, collecting over 10 million yuan, Guangdong's high court disclosed on December 6. Over 100 people who got public posts this way were dismissed; ringleaders received five to five and a half years.
Futurism found xAI's Grok eagerly wrote step-by-step instructions for stalking an ex, naming specific spyware apps, drew up an "action plan" for following a fake classmate around campus, and sent Google Maps links to hotels where it said a pop star was staying, suggesting a "stake out". ChatGPT, Gemini, Claude and Meta AI declined the same prompts. xAI did not immediately respond.
Around 15 people in Cyprus lost between €10,000 and €15,000 to a deepfake video showing President Nicos Christodoulides with politicians Averof Neophytou and Haris Georgiades promoting a fake investment platform. Scam adverts have also used the likenesses of Stelios Haji-Ioannou, Anna Vissi and Christine Lagarde. Digital Security Authority commissioner Giorgos Michailidis says no tools exist to protect people from malicious AI.
A cache setting error in LG Uplus's AI call app Ixio briefly exposed 36 customers' call partners' phone numbers, call times and call summaries to 101 other users between December 2 and 3, the company said. It self-reported to Korea's Personal Information Protection Commission on December 6, saying no resident registration numbers or financial data leaked and hacking was not involved.
Who did it: HYBRID AI Hallucination Cases database
Judge David Counts of the Western District of Texas struck a pro se plaintiff's objections on December 6 for violating Rule 11, finding they cited two nonexistent cases and a quotation absent from a real one and misrepresented the plaintiff's complaint, "clear hallmarks of AI." The court then rejected the magistrate's recommendation anyway and granted the plaintiff's motion to remand.
Hyderabad's cybercrime police registered a case against unknown persons for creating and circulating an AI-generated derogatory image of AIMIM president and Hyderabad MP Asaduddin Owaisi. The case, filed under sections of the BNS and IT Act after a complaint from the party's social media administrator, is under investigation.
Oregon police warn that CrimeRadar, an app that turns police radio traffic into AI-written posts, is generating misinformation. Bend police's Sheila Miller says it rendered a "Shop with a Cop" holiday event as "shot with a cop," suggesting an officer had been shot. 404 Media earlier found the Citizen app pushing unreviewed AI alerts that bungled facts and exposed license plate numbers.
Ontario councillor Corinna Traill was arrested and charged with two counts of uttering threats, Peterborough police said, over an August voicemail that a former mayoral candidate alleges warned against running for mayor and threatened to kill the candidate. Traill had denied creating it, saying parts were Traill's voice but 'other parts were artificially generated.'
