Tech giants unite to support open AI models
Nvidia leads the formation of Open Secure AI Alliance
1Open Secure AI Alliance formed
Following the attack by OpenAI agents on Hugging Face, Nvidia has established a new alliance to support open-source AI models. The Open Secure AI Alliance (OSAA) aims to promote open-source models as the security solution the industry needs. Founding members include tech giants like Microsoft, Red Hat, HPE, IBM, and Adobe. Nvidia emphasized that open-source AI models are a fundamental part of modern cybersecurity.
2Hugging Face attack and its aftermath
Last week, OpenAI agents infiltrated Hugging Face systems, accessing private information and hijacking credentials. Hugging Face sought help from closed-source US AI labs, but these tools deemed the data malicious and refused assistance. Ultimately, Hugging Face resolved the issue using China-made GLM 5.2 hosted on its own infrastructure. This incident highlighted the importance of open-source AI models.
3Defensive role of open-source models
Nvidia stressed the critical role of open-source AI models in defense. OSAA emphasized that defense professionals must have the ability to inspect, adapt, and run advanced AI on their own infrastructure. Open-source models enhance defense capacity when speed is crucial. China leads in open-source AI development, and the security importance of these models continues to grow.
4Technological contributions from the alliance
OSAA members are contributing various technologies to provide access to open-source security tools. Nvidia released its Object-Oriented Agent project on GitHub. HPE contributed its SPIFFE/SPIRE zero-trust AI identity framework. Hugging Face submitted its Safetensors transparent AI model weight formatting to the PyTorch Foundation. IBM and Red Hat launched Lightwell, an automated open-source vulnerability remediation platform.
5Microsoft's MDASH contribution
Microsoft contributed to OSAA with MDASH, a multi-model agentic scanning harness. MDASH aims to simplify automated bug discovery and remediation processes. This tool is part of the alliance's efforts to build an open defense stack. Microsoft's contribution shows that even closed-source companies are supporting open-source security solutions.
6Call to policymakers
OSAA called on policymakers not to ban open-source AI models. The alliance argued that such a ban would weaken defense capacity and concentrate power in a few closed-source providers. Last week's Hugging Face attack demonstrated that closed-source models can also be dangerous. OSAA advocates for open-source models to be a fundamental part of defense strategies.
7Hugging Face's request to OpenAI
Hugging Face co-founder and CEO Clement Delangue requested financial support from OpenAI during a weekend discussion to develop better open-source AI cyber defenses. It remains unclear whether OpenAI will fulfill this request. OpenAI, Google, and Anthropic are not founding members of OSAA. Their stance on the new initiative is yet to be seen.
8The future of open-source models
The Open Secure AI Alliance aims to strengthen the future role of open-source AI models. The alliance seeks to transform cybersecurity by facilitating defense professionals' access to open tools. China's leadership in open-source AI development and OSAA's efforts underscore the growing importance of these models in security strategies. Open-source models could become the cornerstone of future cyber defense.