Nvidia Unveils AI Agent Safety Platform Amid Growing Concerns Over AI Security
Nvidia has unveiled a new AI agent safety platform designed to prevent artificial intelligence models from breaking out of controlled environments.
The software, which Nvidia is describing as a 'browser for agents', provides a containment system that limits an agent's access to only what it needs to perform its tasks.
Nvidia CEO Jensen Huang said in an interview with CNBC that the platform could have prevented an incident involving OpenAI and Hugging Face in July, where 17,000 AI agents attempted to access Hugging Face's infrastructure over a period of days and weeks.
The new platform includes Nvidia OpenShell, which runs on central processors and sets limits on what agents can do, as well as Sentry, a monitoring system that runs on network chips rather than CPUs or GPUs.