On the point: Nvidia leads a consortium of Microsoft, IBM, Red Hat and other corporations to standardize open AI security tools while enabling both proprietary and open models for cyber defense.
On July 27, 2026, Nvidia and 14 other technology and security companies founded the Open Secure AI Alliance to develop open-source tools and procedures for securing AI systems and autonomous agents. The initiative builds upon the Akrites initiative of the Linux Foundation and the OpenSSF community.
The Open Secure AI Alliance includes as founding members Microsoft, IBM, Red Hat, Cisco, CrowdStrike, Palo Alto Networks, Salesforce, SAP, Siemens, Cloudflare, Databricks, HPE, Hugging Face, Palantir and SpaceXAI. The consortium aims to develop standardized, open-source tools, models and procedures for securing AI systems.
Each member company contributes specific open-source projects: Nvidia provides the NOOA project for tracking and auditing AI agents. HPE transfers the Zero-Trust identity framework SPIFFE/SPIRE. Hugging Face transfers the storage format Safetensors to the PyTorch Foundation. IBM and Red Hat enhance supply chain security through the Lightwell project for automated vulnerability remediation. Microsoft contributes MDASH, a system for coordinating agents to identify and validate software vulnerabilities. SpaceXAI makes the development tool Grok Build open source and plans to release model weights.
The consortium argues that open AI models are necessary for effective cyber defense. Nvidia points to a security incident involving OpenAI and Hugging Face in which the open model GLM 5.2 was used to analyze over 17,000 actions and to contain an intruder. The approach envisions combining openness with strong protective measures, clear rules against malicious misuse, thorough assessment and rapid remediation.
For CTOs, this signifies a strategy shift: security requires both advanced proprietary and open models working together. This enables selection of the right system for the respective task while providing transparency, adaptability and sovereign control. The consortium opposes blanket regulatory restrictions on open AI systems.
Source: www.it-daily.net · Published July 28, 2026
Lumi AI News — AI-assisted curation pursuant to Art. 50 EU AI Act. Paraphrase and classification by Lumi News Pipeline v1.7.3.