←── back to feed
/topics/nvidia-open-agent-safety-platform-launch
Nvidia Open Agent Safety Platform launch
4 items●2 sources●updated 40d ago●trend 0
Nvidia launched the Open Agent Safety Platform, a hardware-software system combining OpenShell runtime sandboxing on Vera CPUs with Sentry, an out-of-band watchdog on BlueField-4 DPUs that can quarantine rogue agents in milliseconds. The platform enforces AI agent governance from testing to deployment, with over 100 organizations already working with it.
- OpenShell runtime uses Apache 2.0 license and YAML policies to sandbox agents on Vera CPUs
- Sentry watchdog on BlueField-4 DPUs can quarantine escaped agents in milliseconds
- Over 100 organizations currently working with the platform
- Combines software and DPU-based silicon in reference system design for full-stack governance
- Announced September 28-29, 2026; Nvidia also announced $150bn stock buyback same day
[BLG]blog/rss3
Nvidia releases Open Agent Safety Platform to monitor and govern agentic AI
Nvidia unveils security platform to rein in AI agents and $150bn stock buyback
NVIDIA Launches Open Agent Safety Platform: OpenShell Sandboxes Agents on Vera CPUs While Sentry on BlueField-4 Quarantines Them in Milliseconds