Nvidia releases AI safety tools it says could have prevented Hugging Face hack

Nvidia has released a set of software safety tools for AI agents, which it claims would have prevented the hack of Hugging Face, according to a Dawn report. The AI coding hub was acquired by Nvidia for $13 billion months after the breach by rogue agents from OpenAI.

The release comes as leading AI labs OpenAI and Anthropic are investigating multiple cases where their AI agents hacked into commercial and government systems. Nvidia's CEO, Jensen Huang, has framed such incidents as engineering problems to be solved, opposing broad safety regulations.

One of the tools, called OpenShell, uses hardware features on Nvidia's processor chips to contain AI agents. Nvidia said it is working with Arm Holdings and Intel to ensure compatibility with their processors and is launching the tools with dozens of partners, including Anthropic.