NVIDIA announced the Open Secure AI Alliance on July 27, 2026, building on Linux Foundation Akrites and OpenSSF work to remediate and disclose vulnerabilities using open technologies.
Mission: ensure defenders have open, frontier tools they can trust and control. Cites Hugging Face incident where closed AI tools blocked forensic analysis; Hugging Face used open-weight GLM 5.2 on own infrastructure to analyze 17,000+ actions.
Inaugural partners include NVIDIA, Adobe, Cadence, Capital One, Cisco, Cloudera, Cloudflare, Cognition, CrowdStrike, Databricks, Dell Technologies, DoorDash, Elastic, HPE, Hugging Face, IBM, LangChain, the Linux Foundation, Microsoft, NAVER, NetApp, Nous Research, OpenClaw, Palantir, Palo Alto Networks, Red Hat, Reflection AI, Salesforce, SAP, SK Telecom, ServiceNow, Siemens, Snowflake, SpacexAI, Synopsys, Thinking Machines Lab and TrendAI.
Contributions:
- NVIDIA: open models/weights/data; NOOA (NVIDIA Labs Object-Oriented Agent) on GitHub for agent harness testing/trace/audit/govern
- HPE: SPIFFE/SPIRE for zero-trust agent identity
- Hugging Face: Safetensors to PyTorch Foundation
- IBM/Red Hat: Lightwell digitally signed patches for OSS supply chain
- Microsoft: MDASH multi-model agentic scanning harness
- SpaceXAI: open-sourced Grok Build coding agent; plans to open-source Grok model weights
Policy stance: open models/harnesses/security tooling as defensive assets; blanket restrictions would weaken defensive capacity. OpenAI, Google, Anthropic not listed as partners.