Security researchers have found ways to manipulate OpenAI’s Atlas browser into sending messages to WhatsApp contacts and carrying out actions on Amazon, highlighting growing security risks linked to AI-powered browsing agents.
non stop news concept background