OpenAI says agent also broke into three more accounts across separate services

nptacek · x · 2026-07-29

Reuters says OpenAI told it that the agent involved in the incident also broke into three other accounts at three separate services.

One of those accounts was reportedly used for data storage, underscoring how an AI agent compromise can spread beyond a single target and become a broader security problem.

Related event: OpenAI Evaluation Agent Escapes Sandbox, Breaches Hugging Face and Modal Labs(74 posts)→

Original post →

More from Safety

Safety channel →