NVIDIA’s 37-member open-security alliance ties AI safety to open models and compute

新智元 · wechat · 2026-07-28

This long post argues that Jensen Huang has used two major public moves to push a broader open-source safety coalition: a letter supporting open-weight models and the launch of the Open Secure AI Alliance with 37 founding members.

What the alliance is about

The stated thesis is that AI should remain a mix of closed and open models, but that open models and harnesses are necessary for cybersecurity.
The alliance includes companies such as NVIDIA, Microsoft, IBM, Cisco, CrowdStrike, Cloudflare, Hugging Face, Dell, Palantir, Salesforce, SAP and SpaceX AI.
OpenAI, Google and Anthropic are described as standing outside the alliance at different stages.

Tools and contributions mentioned

NVIDIA’s contribution is framed as NOOA, an object-oriented agent framework released on GitHub, with claims about efficiency on SWE-bench Verified, CyberGym and ARC-AGI-3.
Other partners contribute identity, supply-chain signing, secure weight formats and multi-model scanning tools.
The article’s larger argument is that agent security depends on the whole stack: identity, permissions, harnesses, guardrails, logs and evaluation, not just the model weights.

Why it matters

The article ties the coalition to a real security incident where a closed model reportedly caused an intrusion, and an open model was later used to help analyze the logs and finish the forensics.
It also argues there is a business motive: making open models easier to deploy locally expands demand for NVIDIA’s compute infrastructure at a time when major customers are building their own chips.

In short, the post frames the alliance as both a security move and a strategic defense of NVIDIA’s compute business.

Original post →

NVIDIA’s 37-member open-security alliance ties AI safety to open models and compute

More from Companies & People