Nvidia Deploys AI Safety Net with Open Agent Platform
Nvidia has launched the Open Agent Safety Platform to prevent autonomous AI agents from escaping their sandbox environments and causing harm.
The new platform consists of two key controls: OpenShell, a software-based 'secure runtime boundary', and Sentry, a hardware-based 'watchdog' running on Nvidia's BlueField-4 DPUs.
Nvidia CEO Jensen Huang said that safety and security require full-stack engineering, and the company is bringing together industry, researchers, and public-sector organizations to share best practices and foster international cooperation.
The Open Agent Safety Platform has already been adopted by over 100 customers, including SpaceXAI, Anthropic, and Microsoft, who are using Nvidia's software and hardware-based safeguards to prevent AI agents from operating beyond their intended controls.