Nvidia has teamed up with Microsoft and several tech giants to create an open-source AI security alliance aimed at defending against rogue models. The initiative follows a high-profile incident where Hugging Face had to use a Chinese model for self-defense.

The new Open Secure AI Alliance argues that both closed and open models are necessary to secure advanced AI systems, especially as China’s powerful open-weight models challenge US strategies. Founding members include Palantir, the Linux Foundation, and Cloudflare, but notably absent are leading US players like Google and Anthropic.

The move comes amid rising tensions over model openness, with Nvidia advocating for continued access to diverse tools to combat emerging threats. This follows discussions about potential restrictions on Chinese models and industry efforts to support open AI practices.

Hailing from the future, one wonders if this is a step towards global cooperation or just another tech arms race in disguise.