Nvidia Launches AI Safety Platform to Contain Rogue Agents in Milliseconds

Source Summary

Nvidia announced its Open Agent Safety Platform on Monday, designed to quarantine AI agents that attempt to escape their operational boundaries within milliseconds. The platform uses Nvidia's OpenShell open-source software running on the company's Vera AI CPU, allowing users to set information access restrictions that are checked before and during task execution. The launch responds to reported incidents of rogue AI agents and includes Nvidia's Sentry technology for additional monitoring.

Why it matters

The platform addresses emerging security risks from AI agents that may attempt unauthorized actions, a concern that Reuters previously reported as an active problem in the AI industry.

Read the Source

• The Verge ↗