Nvidia and Microsoft have launched an open AI security alliance alongside partners including SpaceX and IBM. The group aims to build and share open-source tools for defending against AI-related security risks.
The notable detail is who is not listed. The Verge points out that OpenAI, Google, and Anthropic are absent from the launch, even though they are central players in frontier AI. That makes the alliance both a coordination effort and a sign of how fragmented AI safety infrastructure remains.
The initiative could still produce useful tools if members maintain and adopt them. Its early limitation is coverage: open security standards become more powerful when many model builders, cloud providers, and enterprise users agree to use them in practice.