NVIDIA backs open AI security alliance and says defenders need open models too

NVIDIA Blog · rss · 2026-07-27

NVIDIA says it is helping launch the Open Secure AI Alliance, a coalition focused on AI safety and cybersecurity built around open technologies.
The post argues that defenders need open models, open harnesses, and open tooling so they can inspect, adapt, and run security systems on their own infrastructure.
It cites the Hugging Face security incident as an example of why closed tools can hamper forensic response, and says Hugging Face used an open-weight GLM 5.2 model to analyze more than 17,000 actions during containment.
The alliance includes companies such as Microsoft, Cisco, CrowdStrike, Databricks, HPE, IBM, Red Hat, ServiceNow, Snowflake, and others.
NVIDIA also highlights new work like NOOA, plus related open security efforts from HPE, Hugging Face, IBM/Red Hat, Microsoft, and SpaceXAI.
The post ends with a policy appeal: regulators should treat open frontier security tools as defensive assets, not liabilities.

More from Infra