NVIDIA Launches Open Agent Safety Platform to Secure Agents From Testing to Deployment

NVIDIA has launched the NVIDIA Open Agent Safety Platform, an open software platform and reference system design. This initiative aims to strengthen AI security from testing to deployment, providing full-stack governance across software, hardware, compute, and robotics systems. The platform ensures AI agents remain within operator-defined boundaries. Over 100 organizations are already utilizing this technology, including prominent companies such as Anthropic, Microsoft, SAP, Scale AI, and JPMorgan Chase to secure their AI agent deployments.