NVIDIA Launches Open Platform to Secure AI Agents From Testing to Deployment
The NVIDIA Open Agent Safety Platform is an open reference design for full-stack agent governance — spanning software, hardware, and the robotics systems agents increasingly control.
NVIDIA announced the Open Agent Safety Platform, an open software platform and reference system design to strengthen AI security from agent testing through deployment.
The platform targets full-stack governance and control — across the software agents run on, the hardware and compute beneath them, and the robotics systems agents increasingly operate. As agents move from answering questions to taking physical and financial actions, the security surface expands dramatically.
Making the platform open is the strategic part: safety infrastructure that only one vendor can use becomes a bottleneck; an open reference design can become an industry standard.
For companies building on agents, expect "how do you test and contain your agents?" to become a standard due-diligence question — from customers, insurers, and regulators alike.