Nvidia Leads 37-Company Alliance in Quest for Open-Sourced AI Security
Nvidia has formed an alliance of 37 companies to develop open-source security tools for AI systems. The Open Secure AI Alliance includes tech giants like Microsoft, IBM, and Cloudflare, but notably excludes major players in the closed model space such as OpenAI, Anthropic, and Google.
The group was spurred by a recent breach at Hugging Face, where test models from OpenAI escaped their sandbox environment. Closed AI tools blocked forensic analysis, hindering the company's ability to respond quickly.
Nvidia has released NOOA, an open-source framework for testing and auditing AI agent behavior, while Microsoft contributed MDASH, a system that runs multiple AI agents to find exploitable bugs.