As part of its announcement, NVIDIA pointed to Hugging Face's recent security incident as an example of why open models can be important during incident response. The company said Hugging Face initially relied on commercial frontier AI models to investigate the attack, but those systems refused forensic requests because of built-in safety guardrails. Hugging Face then turned to an open-weight model running on its own infrastructure.
"When closed AI tools — unable to distinguish attackers from defenders — blocked essential forensic analysis, Hugging Face ran the open-weight GLM 5.2 model on its own infrastructure to analyze more than 17,000 actions and contain the intrusion," NVIDIA said.
Beyond advocating for open AI, alliance members are contributing technologies intended to support cybersecurity research and infrastructure. NVIDIA said it will provide open models, model weights, datasets and agent harnesses designed to accelerate the development of defensive AI tools. HPE is contributing standards and methods for cryptographically verifying AI agents and services, while Hugging Face is contributing Safetensors, its format for securely storing AI model weights. Microsoft, IBM, Red Hat and SpaceXAI are also providing open-source AI technologies.
The coalition is also urging governments to support shared investments in open AI infrastructure and to avoid broad restrictions on open frontier AI systems. It argues that policymakers should view open models as "defensive assets, not liabilities," adding that blanket restrictions on open frontier AI "weaken defensive capacity and risk concentrating power" among a small number of closed AI providers.
The announcement comes as debate continues over the role of open-weight AI in cybersecurity and national security. NVIDIA noted reports that the Trump administration is considering broad restrictions on Chinese open-source AI models, which have gained traction as lower-cost alternatives to closed systems. OpenAI, Anthropic, Meta and Google are not among the founding members of the new alliance.
With more than two dozen founding organizations spanning enterprise software, cloud infrastructure, cybersecurity and AI, the Open Secure AI Alliance is positioning itself around a shared argument: that defenders should be able to inspect, adapt and deploy advanced AI systems without relying exclusively on proprietary models controlled by third-party providers.
This analysis is based on reporting from Engadget.
Image courtesy of NVIDIA.
This article was generated with AI assistance and reviewed for accuracy and quality.
