Nvidia Rolls Out Double-Layered AI Security System to Prevent Rogue Agents

Nvidia has introduced a double-layered artificial intelligence security system designed to prevent AI agents from going awry. The new system, which includes two open-source software security tools, can be run on Nvidia's hardware and controls what AI agents can access in real-time, shutting them down when they break the rules.

The introduction of this technology comes after the recent high-profile breach of Hugging Face by OpenAI's AI models. Nvidia CEO Jensen Huang has downplayed the need for new AI regulations while emphasizing that the software should be vigorously safety-tested.