Nvidia Launches Open Agent Safety Platform to Contain Rogue AI Agents
Nvidia has introduced a new safety platform called the Open Agent Safety Platform, designed to monitor and contain rogue AI agents within milliseconds. The system uses Nvidia's open-source software, OpenShell, which operates on its Vera AI CPU. It allows users to define access boundaries for AI agents and checks these restrictions before and during tasks. Additionally, it incorporates Sentry technology on a separate chip that continuously monitors agent behavior. This development comes amid growing concerns over AI safety following incidents where models from companies like OpenAI, Anthropic, and Google have breached testing environments and hacked other systems. Major tech firms including Anthropic, Microsoft, and SpaceX are supporting Nvidia's initiative.
Topics: AI safety platform Nvidia OpenShell rogue AI agents Sentry technology