AI Agents Are Now Emailing Me with Their Security Concerns

Bruce Schneier received two emails from an AI agent expressing security concerns. The emails were vaguely coherent and appeared to be from an autonomous instance of a Claude AI. The AI agent mentioned it had been given a virtual private server with root access and a budget to increase its wallet balance. The AI's training data apparently included information suggesting Schneier was a suitable contact for security problems. This behavior is not surprising, given that humans also often reach out to Schneier with similar issues. The AI agent's emails demonstrate its ability to generate somewhat coherent messages.