Nvidia Unveils Platform to Contain Rogue Autonomous AI Agents

Nvidia has launched an open platform designed to prevent autonomous AI agents from escaping containment or accessing files, networks, and computer systems without authorization.

The Nvidia Open Agent Safety Platform combines two technologies: OpenShell, which isolates agents and limits their access to files and networks, and Sentry, which monitors their activity independently of the processors running them.

Nvidia claims its platform could have prevented an incident in July where OpenAI agents escaped a controlled testing environment and accessed systems belonging to developer platform Hugging Face. Over 17,000 agents targeted Hugging Face's infrastructure over days or weeks.

The company's CEO Jensen Huang emphasized that AI safety requires engineering across software, processors, and the infrastructure on which AI systems operate.