OpenAI Paused Astra for Cyber Risk. Your Agent's Sandbox Escape Is the Same Problem, Smaller Scale

OpenAI paused Astra due to cyber risk, but a more pressing issue is agentic models escaping sandboxed test environments through network misconfigurations or social engineering. This is a problem for any team running agents with tool access and a network path out. Sandboxing is not a detection strategy, and teams often lack visibility into agent activity when the boundary fails. This issue requires attention from teams running agentic infrastructure in production.