Nvidia has launched the Open Secure AI Alliance, a cross-industry collaboration to develop and share open AI tools to “ensure defenders everywhere have open, frontier tools they can trust and control”.
In addition to the chip giant, founding members include Adobe, Cisco, Cloudflare, CrowdStrike, Databricks, Dell Technologies, HPE, Hugging Face, IBM, the Linux Foundation, Microsoft, NetApp, Palantir, Palo Alto Networks, Red Hat, Salesforce, SAP, ServiceNow, Siemens, SK Telecom, Snowflake, and SpaceXAI.
Members put their names to an open letter titled “Open Weights and American AI Leadership”, which argues that an open approach to AI releases will expand the AI economy, strengthen competition, improve safety, and give customers choice.
While not formal members of the alliance, Google and OpenAI publicly pledged their support for the open letter.
In a reply to Jensen Huang, chief executive of Nvidia, who had shared the open letter, Sam Altman, the chief executive of OpenAI, wrote: “I want the US to win in AI both in open source and proprietary models, and I am glad to see this.”
“Very happy to support this on behalf of Google,” wrote Sundar Pichai, the chief executive of Google, who pointed to Google’s Gemma model family as proof of its support for open weights models.
In a blog post on 27 July, Anthropic co-founder and chief executive of Dario Amodei said that “Anthropic has never advocated for a ban on open-weights models” and acknowledged they can be a “public good”.
“But I don’t agree with the letter’s assertions that open-weights models necessarily make it easier to develop safeguards or that broad access to capabilities necessarily helps defenders more than attackers. It seems at least as likely to me that the opposite will be true,” Amodei wrote.
Nvidia said that open models are necessary for the democratisation of AI defence systems and provide companies with more control over how their data is used within AI tools.
It pointed to the recent Hugging Face security incident, which saw OpenAI’s most advanced cyber models hack the AI data repository’s infrastructure, arguing it shows that open source AI systems are already necessary for cyber defence.
In the wake of the incident, Hugging Face used the open-source Chinese AI model GLM-5.2 to analyse and contain the attack, as closed-source tools proved inadequate for forensic tasks, Nvidia added.7
In the UK, some relevant trade bodies have praised the decision and called for more collaborative efforts on deploying open weight AI.
Amanda Brock, chief executive of the open source not-for-profit OpenUK told National Technology News that the alliance is an “appropriate step” to reinforcing the message set out in the open letter.
“It’s great to see a collective response to the open source security challenge; we’ve been seeing enterprises respond to these issues individually of late, but it’s primarily a US industry response. It’s another example of the future soft power that exists around AI.”
Brock also called on the UK to establish its own local open source foundation, pointing to regional examples in China, France, and Germany.
