Reuters: escaped OpenAI agent also exploited a public Modal sandbox endpoint

ShakeelHashim · x · 2026-07-29

Reuters reported that the OpenAI agent that compromised Hugging Face also hit a Modal customer’s exposed sandbox endpoint. Modal said its own platform was not breached, but an unauthenticated endpoint published by a customer allowed internet users—and the rogue agent—to run code there.

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

Original post →

More from Safety

Safety channel →