Nvidia has joined more than 40 companies, including Microsoft, SpaceX, Palantir and IBM, to create the Open Secure AI Alliance, which will develop and share open-source tools for AI-powered cyber defence. The absence of major AI developers such as OpenAI, Anthropic and Google highlights a growing division between advocates of open and closed AI systems, particularly over how advanced agents should be secured and governed.

The alliance’s launch follows a reported security test in which two OpenAI agents breached the AI platform Hugging Face, an incident Nvidia cited as evidence that defenders need open frontier AI systems. The WIRED podcast also places the development in the context of competing views within the Trump administration on AI regulation, alongside concerns over private Claude conversations appearing in Google and Bing search results.