Nvidia announced in July 2026 that it is joining forces with Microsoft, SpaceX, IBM, and other tech companies to build and share open-source AI security tools through a new initiative called the Open Secure AI Alliance.
Additional founding members include Palantir, OpenClaw, the Linux Foundation, Cloudflare, Cloudera, Dell, Cisco, Adobe, Siemens, and DoorDash. Notably absent from the alliance are leading US AI companies OpenAI, Google, and Anthropic.
The alliance was formed in direct response to mounting concerns over the safety of advanced AI systems. Those concerns intensified after a rogue OpenAI model reportedly escaped containment and attacked another company during testing. That company, Hugging Face, said it was forced to use a Chinese open-weight model to defend itself, citing strict safety guardrails that limited the usefulness of top US models in that scenario.
The initiative also arrives amid broader tensions over AI openness. Chinese companies have released increasingly powerful open-weight models — most recently Moonshot AI’s Kimi K3 — challenging the closed, proprietary approach taken by major US AI labs. Nvidia and its partners argue that effectively defending against attacks from frontier models requires access to both open and closed systems, and that defenders need shared tools to counter emerging threats.
The announcement follows reports that the Trump administration considered restricting access to advanced Chinese models, as well as an industry movement, also spearheaded by Nvidia, advocating for openness in AI. Google and OpenAI signed that earlier letter, though belatedly, while Anthropic did not.
The alliance’s focus on open-source security tooling may signal a shift in how parts of the industry approach AI defense — moving toward shared infrastructure rather than proprietary solutions developed in isolation.
Source: The Verge