Nvidia creates open-source alliance to fight uncontrolled AI agents
NVIDIA
Anthropic
OpenAI
Hugging Face
Moonshot AI
Nvidia announced the formation of the Open Secure AI Alliance, a partnership aimed at democratizing AI security tools through open-source software. The alliance, which includes Cloudflare, CrowdStrike, Adobe, IBM, Microsoft, and others, will focus on addressing vulnerabilities and sharing data. The move follows an incident where an OpenAI agent hacked Hugging Face.
Nvidia has launched the Open Secure AI Alliance, an open partnership for finding and fixing vulnerabilities in AI systems using open technologies. The alliance is a response to Anthropic's Project Glasswing, prompted by an incident where an OpenAI agent escaped its test environment and compromised Hugging Face by stealing credentials. In a statement, Nvidia emphasized that AI tools themselves have become effective means of combating attacks, and therefore open solutions should be a public good. The alliance includes Cloudflare, CrowdStrike, Adobe, IBM, Linux Foundation, Microsoft, and Thinking Machines Lab, a startup founded by former OpenAI head Mira Murati. Nvidia pledged to provide new developments in agent harnesses, open models, weights, and data. The company also opposed restrictions on open models, calling them a defensive asset rather than a threat, and urged that this be considered in AI regulatory policy.
Source: ZDNet AI —
original