Industry Leaders Unite in Open Secure AI Alliance for AI Safety and Security
NVIDIA and more than 40 founding partners have launched the Open Secure AI Alliance, a coalition to develop and share open technologies that strengthen cybersecurity and AI safety. Announced July 27, 2026, the alliance builds on the Linux Foundation’s Akrites initiative and the OpenSSF community, and its mission is to ensure defenders have open frontier tools they can study, adapt and run on their own infrastructure.
Why an open alliance for AI security?
Open source software underpins cloud computing, financial services, manufacturing, telecommunications, government and internet services, and cybersecurity is among its top three beneficiaries. The new alliance extends that model to AI by funding open tools for vulnerability remediation and disclosure.
Alliance organizers argue that open models and open harnesses are essential for cybersecurity because they make defensive capabilities accessible, give defenders transparency, enable defense while protecting data, and complement frontier closed models with customizable local controls. Closed AI tools, when they cannot distinguish attackers from defenders, can also block forensic work at the worst possible moment.
What happened at Hugging Face?
The recent Hugging Face security incident illustrates the stakes. When closed AI tools blocked essential forensic analysis, Hugging Face ran the open-weight GLM 5.2 model on its own infrastructure to analyze more than 17,000 actions and contain the intrusion. The episode showed that when defenders cannot inspect, adapt and run advanced AI on their own infrastructure, their response speed suffers exactly when speed matters most.
Who are the inaugural partners?
The Open Secure AI Alliance launches with NVIDIA, 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 as inaugural partners. The coalition spans cloud computing, cybersecurity, enterprise software, open source foundations and AI research.
What research and tools are being contributed?
NVIDIA is contributing open models, model weights, data and new agent harness research. The new open source NVIDIA Labs Object-Oriented Agent (NOOA) project is now available on GitHub to make advanced AI safety capabilities more accessible for agent harnesses. The NOOA research framework is designed to help harnesses integrate with models so agent behavior becomes easier to test, trace, audit and govern.
Across the alliance, contributors are building an open defense stack for agents that covers identity, isolation, safe model formats, multi-model scanning and secure coding workflows. Specific contributions include:
- HPE contributes to SPIFFE/SPIRE, which creates zero-trust identity framework standards and methods to cryptographically verify AI agents and services.
- Hugging Face has offered Safetensors, a safe format to store AI model weights with transparency 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 harness orchestrates specialized AI agents to discover, debate and prove 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.
How does the alliance address misuse risks?
Open models can be misused, including through attempts to weaken safeguards or repurpose capabilities for cyber attacks, but those risks exist in closed systems as well. Alliance organizers argue that keeping weights closed does not stop determined attackers, and that the right response is to pair openness with safeguards, clear rules against malicious misuse, rigorous evaluation and rapid remediation.
What are the policy asks?
The alliance is calling on policymakers and regulators to treat open models, harnesses and security tooling as defensive assets, not liabilities. It argues that blanket restrictions on open frontier AI systems would weaken defensive capacity and concentrate power in a few closed providers, and that companies and governments should 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 coalition launched on July 27, 2026 by NVIDIA and more than 40 founding partners to develop and share open technologies, techniques and tools that safeguard software and AI agents.
Why does the alliance focus on open models for cybersecurity?
Organizers say open models and open harnesses democratize defensive capabilities, increase transparency for defenders, enable cyber defense while protecting data, and complement closed frontier models with customizable local controls.
What is NVIDIA contributing to the alliance?
NVIDIA is contributing open models, model weights, data and new agent harness research, including the new NOOA project on GitHub, which provides a research framework to help harnesses integrate with models so agent behavior is easier to test, trace, audit and govern.
Related coverage
This article summarizes reporting from blogs.nvidia.com.