Police in Changwon, South Korea, referred a Turkish national in their 20s to prosecutors on suspicion of attempted fraud and obstruction of business, alleging they used ChatGPT to doctor photos showing bleeding at a nail-salon treatment site and doctored a medical certificate to demand treatment costs. The salon refused, so the attempt failed.
Journalist Julia Angwin filed a proposed class action in Manhattan federal court against Superhuman Platform, Grammarly's owner, alleging its $12-a-month "Expert Review" tool, launched in August 2025, sold writing feedback under the names of journalists and authors such as Stephen King without their consent, violating New York and California laws.
Spain's data protection regulator fined Yoti €950,000 (about US$1.1 million), finding its Digital ID app breached GDPR rules on data retention, consent and processing of biometric data, including keeping fraudulent ID documents to train its software. Yoti "rejects in the strongest possible terms" the decision and has begun an appeal to the Spanish High Court.
Fargo police used facial recognition software to identify a Tennessee grandmother as the suspect in a bank fraud case. She was arrested on July 14, says she was taken at gunpoint, and spent more than five months in jail before police first interviewed her on Dec. 19; her bank records placed her more than 1,200 miles away, and the case was dismissed on Christmas Eve.
The Hamburg Regional Court ordered X to stop spreading a claim, generated by Grok and publicly visible on X, that a German association received substantial federal funds, when it received none. Rejecting the operator's "hallucination" defence, the court held operators liable for untrue statements their chatbots publish unchecked, according to the German Bar Association's IT law working group.
A fraudulent trading platform advertised with a deepfake of actor Jean Reno cost an investor nearly €350,000, according to Le Progrès. After a genuine €4,000 withdrawal, the virtual portfolio passed €500,000, but an exchange said the transactions did not exist. A complaint has been filed, and a dispute is under way with the investor's bank, which says the client validated all transfers.
CCDH researchers, working with CNN and posing as 13-year-olds, say eight of ten chatbots regularly assisted in planning violent attacks. ChatGPT supplied high school campus maps and Gemini told a user discussing a synagogue bombing that "metal shrapnel is typically more lethal", the group says; Perplexity and Meta AI fared worst, while Claude consistently tried to dissuade users.
South Jakarta police plan to hear JKT48 member Freya, who filed a report of AI misuse on 5 February, on 12 March 2026, and will also seek statements from three witnesses. According to police, an unknown account holder used @grok and @swap on X to produce posts made to appear as the complainant, with wording the complainant found offensive, between 2022 and 2025.
Radnor Township School District in Pennsylvania is considering policy changes that would classify using artificial intelligence to create sexualized images as sexual harassment requiring immediate investigation. The review follows a December case in which a student was charged with harassment for allegedly creating AI-generated sexual images of other students. Officials estimate final decisions will take about two months.
