Why NVIDIA Launched the Open Agent Safety Platform
With Anthropic, Salesforce and other tech leaders embedding NVIDIA controls, CEO Jensen Huang describes the platform as “the foundation of the AI economy”
TL;DR
- NVIDIA's Open Agent Safety Platform aims to secure AI agents by embedding controls into the underlying infrastructure.
- The platform includes NVIDIA OpenShell to establish safe boundaries around AI agents and NVIDIA Sentry for hardware-based security enforcement.
- Sentry can quarantine agents in milliseconds if they attempt to exceed authorized software boundaries.
- More than 100 companies, researchers, and public-sector organizations are adopting the platform.
- Collaborators include Anthropic, Salesforce, SpaceXAI, Scale AI, Cisco, Microsoft, and major players in finance and energy.
- CEO Jensen Huang describes the platform as "the foundation of the AI economy" and a "trust layer for safe agent systems."