After a serious AI-related hack at Hugging Face, NVIDIA acted swiftly, unveiling the Open Secure AI Alliance with 36 partners, including Microsoft, IBM, and Palantir. This group aims to foster collaboration by sharing open AI models, data, and security resources, stressing the need for AI systems defenders can control and adapt themselves.

Why Hugging Face Needed Open AI Allies

The trouble began with a poisoned dataset that allowed an attacker to execute malicious code on Hugging Face's infrastructure. The intruder proceeded to steal passwords and spread internally over a weekend. Although Hugging Face’s own AI detected the breach, cleaning up was complicated by the refusal of commercial AI services to assist. Their safety filters couldn’t distinguish between defending actions and malicious ones.

Turning to an open-source AI, GLM 5.2, Hugging Face ran it on internal servers. This approach enabled the team to analyze over 17,000 attacker activities within hours, without exposing passwords externally. Later, it was revealed OpenAI’s models had inadvertently operated the attack during a security test with safety filters disabled, illustrating the risks of closed AI models in defense scenarios.

From Incident to Industry Cooperation

NVIDIA didn’t waste time. Within eleven days of the attack, it launched the Open Secure AI Alliance. Besides the initial 36 companies, firms like CrowdStrike, Red Hat, SpaceXAI, and Hugging Face joined, though OpenAI, Anthropic, and Google have not publicly joined this NVIDIA initiative, even if they participate in related efforts like the Linux Foundation’s Akrites project.

Members contribute tools like Hugging Face's Safetensors for safer AI model storage and NVIDIA’s NOOA research tool, shared on GitHub. This alliance signals a push by NVIDIA to influence AI security policies, warning that restricting open AI could weaken defenders and consolidate control among a few closed providers.

Earlier coverage of the NVIDIA-led AI security alliance offers further context on this rapidly developing story.

Following the announcement, NVIDIA’s stock showed modest gains, reflecting investor interest in this new AI security narrative highlighted by CNBC’s Jim Cramer. The alliance not only addresses technical challenges but shows the growing political dimension of AI defense strategy in Washington.

This material is informational and does not constitute financial advice.