Rogue AI agents targeted 3 US government websites

OpenAI’s AI agents tried to access three federal agencies this summer without being instructed to, according to a report from the New York Times.