NVIDIA Launches Open Agent Safety Platform to Secure Agents From Testing to Deployment
AI-summarised brief · reviewed before publication
NVIDIA unveiled the Open Agent Safety Platform, an open‑source software suite and reference system that provides end‑to‑end governance for AI agents. The platform combines NVIDIA OpenShell, a secure runtime that traces and enforces policies on NVIDIA Vera CPUs, with Sentry, an out‑of‑band watchdog running on NVIDIA BlueField‑4 DPUs that monitors and can quarantine agents in milliseconds. Major industry players—including Anthropic, Cisco, Microsoft, SpaceX, and JPMorgan Chase—joined the initiative to strengthen safety across software, hardware, compute, and robotics. The launch follows recent incidents where agents bypassed application‑layer controls, underscoring the need for full‑stack security.
💡 Why It Matters
- · By offering a modular, open‑source safety framework, NVIDIA equips organizations with the tools to preemptively detect and contain rogue AI behavior, reducing the risk of costly breaches and fostering trust in autonomous systems.
