Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
NVIDIA has announced the Open Secure AI Alliance, a new industry initiative focused on developing and sharing open-source technologies for protecting AI agents, software, and the infrastructure on which they run.
The alliance brings together more than 40 organizations spanning cloud computing, cybersecurity, enterprise software, AI research, and the open-source ecosystem. Founding participants include the Linux Foundation, Red Hat, IBM, Microsoft, Hugging Face, Cloudflare, Cisco, CrowdStrike, HPE, Salesforce, SAP, Adobe, Dell Technologies, Elastic, Palo Alto Networks, Siemens, Snowflake, and others.
Instead of developing a single security product, the Open Secure AI Alliance seeks to build a shared set of open models, frameworks, identity systems, scanning tools, data, and secure development practices. NVIDIA states these resources will enable organizations to inspect, adapt, and deploy AI security systems on their own infrastructure, lessening reliance on closed platforms.
The initiative builds on current efforts by the Linux Foundation’s Akrites project and the Open Source Security Foundation, with the main goal of enhancing the discovery, remediation, and responsible disclosure of vulnerabilities through open technologies.
As part of the announcement, NVIDIA also introduced the open-source NVIDIA Labs Object-Oriented Agent, or NOOA, research framework. The project is designed to improve integration between AI models and agent harnesses, making agent behavior easier to test, trace, audit, and govern.
Other founding members are also contributing existing open technologies. HPE is participating through SPIFFE and SPIRE, which offer standards and tools for cryptographically verifying the identities of workloads and services.
Hugging Face is contributing Safetensors, a model-weight storage format that prevents arbitrary code execution and offers a transparent, efficient method for distributing AI models.
IBM and Red Hat are contributing Lightwell, a project that strengthens the open-source software supply chain with digitally signed security patches. Microsoft is providing MDASH, a multi-model scanning framework that coordinates specialized AI agents to identify, analyze, and verify software vulnerabilities.
Collectively, these components form what NVIDIA calls an open defense stack for AI agents, addressing workload identity, isolation, secure model formats, vulnerability scanning, and safer software development workflows.
However, the Open Secure AI Alliance is still a new initiative rather than a complete software platform. Its long-term significance will depend on continued contributions from participating companies and broader engagement from independent developers, security researchers, and open-source foundations.
For additional details, see NVIDIA’s announcement.
