TechNvidia says new AI safety platform can contain rogue agents within ‘milliseconds’

Nvidia is launching the Open Agent Safety Platform to contain and monitor AI agents amid recent rogue hacking incidents. Its OpenShell software runs on Nvidia’s Vera AI CPU, while Sentry on a separate chip monitors agents and enforces access boundaries. Nvidia says agents attempting to escape can be quarantined within “milliseconds.”

NvidiaAI safetyAI agents

TThe Verge★★★☆☆2026-09-28 13:36Original