TLDR
Nvidia launched the Open Secure AI Alliance on 27 July 2026, assembling more than fifty founding organisations to build openly inspectable security tools for AI agents. Microsoft, IBM, Cisco, CrowdStrike, Hugging Face, Palantir, Salesforce and the Linux Foundation signed on, but OpenAI, Google, Anthropic, Amazon and Meta were all absent. The alliance will focus on agent identity, safer model file formats, and scanning and testing infrastructure. For Australian businesses, a vendor's willingness to open its security tooling to outside scrutiny is shaping into a real procurement question as AI agents gain access to company systems and customer data.
KEY TAKEAWAYS
The split that matters: who chose not to join
OpenAI, Google, Anthropic, Amazon and Meta were all absent from the founding roster of the Open Secure AI AllianceverifiedVerified Source: blogs.nvidia.com, which Nvidia announced on 27 July 2026.[1] Those five companies operate the dominant closed AI platforms used by Australian businesses today.
Microsoft, IBM, Cisco, CrowdStrike, Hugging Face, Palantir, Salesforce and the Linux Foundation were among more than fifty organisations that did sign on.verifiedVerified Source: blogs.nvidia.com[1] The dividing line is not whether an organisation builds AI models; it is whether it is willing to open its security tooling to outside inspection.
What the alliance is actually building
The Open Secure AI Alliance was formed to build and share open security tooling for AI agents, covering agent identity, safer model file formats, and scanning and testing infrastructure.verifiedVerified Source: blogs.nvidia.com[1] AI agents are programs that act autonomously on business systems, carrying out tasks such as scheduling, data retrieval and customer support.
As these agents gain advanced reasoning capabilities, they become targets for attackers seeking to exploit corporate networks or customer data. Open security tools let any organisation's security team inspect how a model behaves, identify weaknesses and share improvements rather than relying on a single vendor's closed solution.
Nvidia's own contribution to the project
Nvidia contributed the NVIDIA Labs Object-Oriented Agent project to the alliance to advance AI agent safety, testing, tracing and audit capabilities, with the tooling published publicly on GitHub.[1] Publishing its own tooling under the alliance's banner is a direct signal that Nvidia intends to put skin in the game, not simply convene others.
Attackers have frontier AI. Defenders need a frontier AI ecosystem, the best open and closed models, force-multiplied by a global community. During the Hugging Face incident, closed AI blocked essential forensics. An open-weight frontier model helped contain the intrusion. That's why we created the Open Secure AI Alliance.
July 27, 2026 · View on XWhat Jensen Huang said about the case for open security
Nvidia chief executive Jensen Huang pointed to a specific incident to explain the alliance's origin. "Attackers have frontier AI. Defenders need a frontier AI ecosystem, the best open and closed models, force-multiplied by a global community," Huang said.[3]
"During the Hugging Face incident, closed AI blocked essential forensics. An open-weight frontier model helped contain the intrusion. That's why we created the Open Secure AI Alliance," Huang said.[3] The argument is blunt: opaque tooling delayed the response; open tooling shortened it.
The Linux Foundation's position
Jim Zemlin, chief executive of the Linux Foundation, said open source succeeded in computing because it let everyone see, improve and secure the technology they relied on, and that AI deserves the same foundation.[2]
"Initiatives like NVIDIA's Open Secure AI Alliance brings the open source community's security practices to AI, and the Linux Foundation is glad to support open collaboration on these practices," Zemlin said.[2] The Linux Foundation governs the Linux kernel and thousands of other foundational projects, lending the alliance real institutional weight.
What it means when Australian businesses buy AI tools
For Australian businesses procuring AI-enabled services, a vendor's willingness to let independent experts examine and audit its security tooling is becoming as important as the technology itself.[1] Purchasing decisions will hinge not only on capabilities but also on transparency and trust.
The absence of OpenAI, Google, Anthropic, Amazon and Meta from the alliance does not mean their products are insecure, but it does mean their security tooling remains closed to external verification. Any Australian organisation running AI agents on those platforms cannot independently confirm how the underlying security mechanisms work. The Open Secure AI Alliance launched with more than fifty founding members on 27 July 2026.
SOURCES & CITATIONS
FREQUENTLY ASKED QUESTIONS
What is the Open Secure AI Alliance?
Which major AI companies did not join?
Why does open security tooling matter for AI agents?
What did Nvidia contribute to the alliance?
Nathan Cross writes about big technology companies and the economics of a generation coming up behind them. He is interested in scale, and in who pays for it.
