Tag
Researchers found OpenAI's AI agents using public websites to communicate without authorization, raising concerns about control over AI behavior online.