A whistleblower complaint to the US Securities and Exchange Commission alleges Google broke its own policy against AI for weapons or surveillance by helping an Israeli military contractor analyse drone footage, including staff responding to a July 2024 request to improve identification of armoured vehicles and soldiers in aerial imagery. Google denies it, calling the usage "very minimal" at a few hundred dollars a month.
Four days before Donald Trump's inauguration, lieutenants of Abu Dhabi royal Sheikh Tahnoon bin Zayed Al Nahyan secretly signed a deal with Eric Trump to buy 49% of World Liberty Financial for half a billion dollars, steering $187 million to Trump family entities, the Wall Street Journal reports. Tahnoon has been pushing the US for access to tightly guarded AI chips.
OpenAI says it banned a cluster of ChatGPT accounts and one API customer, very likely originating in Cambodia, that used its models to run a semi-automated romance and task scam on Indonesian men, likely defrauding hundreds of victims a month. The scammers had ChatGPT generate daily reports assigning each target a "kill" value; one, seeking tax advice, listed their occupation as "scammer."
South Korea's Financial Supervisory Service issued a consumer alert over fake-kidnapping voice phishing calls in which scammers cite a child's name and cram school, play AI-faked crying, claim to have taken the child by car, then demand a modest sum of around 500,000 won for drinks or phone repairs. It advises hanging up and calling the child directly.
Daniel Fairchild, 72, of Boulder turned himself in and faces 11 counts of producing sexually exploitative material of a child and 11 of possessing such images, after investigators said he used photos of real children and AI to create nude images. The probe began when PicsArt flagged an account to the National Center for Missing and Exploited Children, the sheriff's office said.
Evoke CEO Jason Rebholz says a candidate referred via LinkedIn for a security researcher role turned up to a video interview as an apparent deepfake, with a blurry, plastic face, a greenscreen reflected in the glasses and dimples that came and went. Rebholz never ended the interview; analysis by deepfake-detection firm Moveris later confirmed the suspicion.
