Open Secure AI Alliance launches with NVIDIA and 30+ partners

Free Instant SEO Audit

By NewSunSEO July 28, 2026

NVIDIA, together with more than 30 industry partners, has formed the Open Secure AI Alliance, a multi-vendor effort to build and share open models and security tools that defenders can inspect, adapt, and run on their own infrastructure. The alliance, announced on July 27, 2026, builds on the Linux Foundation’s Akrites initiative and the OpenSSF community, and it positions open-weight systems as a defensive asset, not a liability, in AI and cybersecurity policy.

Why the alliance exists

The founding statement frames AI security as a choice between defenses locked inside a few opaque systems and defenses built on open models, tooling, and software that any defender can study, adapt, and deploy. Open source software already underpins cloud computing, financial services, manufacturing, telecommunications, government, and internet services, and cybersecurity ranks among the top three beneficiaries. The alliance is designed to remediate and disclose vulnerabilities using those same open technologies.

The alliance argues that closed AI tools cannot tell attackers from defenders at the moment speed matters most. To support the point, the post cites a recent security incident at Hugging Face, in which closed AI tools blocked essential forensic analysis. Hugging Face responded by running the open-weight GLM 5.2 model on its own infrastructure to analyze more than 17,000 actions and contain the intrusion. The post treats that episode as a working demonstration of why open frontier defensive tooling matters for self-defense.

What the alliance will build

According to NVIDIA, the alliance will assemble an open defense stack for AI agents, covering identity, isolation, tooling, guardrails, logs, and evaluation. NVIDIA is contributing open models, model weights, data, and new research on the framework for running AI agents, along with the new NVIDIA Labs Object-Oriented Agent (NOOA) project, now available on GitHub, which the post describes as a research framework that lets these frameworks integrate with models so agent behavior is easier to test, trace, audit, and govern.

Contributions from founding partners

HPE contributes to SPIFFE/SPIRE, a zero-trust identity framework that cryptographically verifies AI agents and services so only authorized workloads communicate and access enterprise resources.
Hugging Face has offered Safetensors, a safe format for storing AI model weights with transparency guarantees and no remote code execution, to the PyTorch Foundation.
IBM and Red Hat’s Lightwell extends security across the open source supply chain with digitally signed patches.
Microsoft’s MDASH multi-model agentic scanning tool coordinates specialized AI agents to find, assess, and confirm exploitable bugs.
SpaceXAI has open sourced the Grok Build terminal-based AI coding agent and plans to open source the weights of the Grok line of models.

The post argues that AI safety and security depend on the full agent stack, not on whether model weights are open or closed.

Who is involved

Founding partners span cloud computing, cybersecurity, enterprise software, open source foundations, and AI research. The post names 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, ServiceNow, Siemens, SK Telecom, Snowflake, SpacexAI, Synopsys, Thinking Machines Lab, and TrendAI, alongside NVIDIA.

The argument on open vs. closed models

The founding post addresses the claim that open models are inherently less safe because they can be misused or modified to remove guardrails. It concedes those risks are real but argues they do not disappear in closed systems, and that keeping weights closed does not prevent determined attackers from seeking or exploiting powerful AI. Instead, the alliance pairs openness with safeguards, clear rules against malicious misuse, rigorous evaluation, and rapid remediation, and argues defenders need both frontier closed models and frontier open models working together.

Policy stance

The post asks policymakers and regulators to treat open models, and security tooling as defensive assets in AI and cybersecurity policy. It warns that blanket restrictions on open frontier AI systems would weaken defensive capacity and concentrate power, dependence, and vulnerability in a few closed providers, and it asks companies and governments to invest in shared open infrastructure for AI defense, including datasets, evaluation frameworks, attack simulators, and red-teaming tools.

FAQ

What is the Open Secure AI Alliance?

The Open Secure AI Alliance is a multi-vendor initiative launched on July 27, 2026 by NVIDIA and more than 30 founding partners to develop and share open models, testing frameworks, and security tools that defenders can inspect, adapt, and run on their own infrastructure.

What is NVIDIA contributing to the alliance?

NVIDIA is contributing open models, model weights, data, and new research on agent testing infrastructure, including the NOOA project on GitHub, which provides a research framework that combines testing tools with models to test, trace, audit, and govern agent behavior.

Why does the alliance argue defenders need open models?

The alliance argues that cyber defenders need open frontier models because they can be studied, adapted, and deployed on the defender’s own infrastructure, avoiding single points of failure. It points to a Hugging Face security incident in which closed AI tools blocked forensic analysis and Hugging Face used the open-weight GLM 5.2 model to analyze more than 17,000 actions and contain the intrusion.