Nvidia launched the Open Secure AI Alliance to improve safety and security for open-weight artificial intelligence models.
Nvidia launched the Open Secure AI Alliance on Monday to address security vulnerabilities in open-weight artificial intelligence models. The initiative, formed in partnership with Microsoft, SpaceX, and Palantir, aims to build and share open AI tools to ensure cyber defenders can inspect and adapt systems for self-defense. The alliance was formed in response to a recent security incident where the startup Hugging Face was unable to use leading U.S. frontier models to defend itself against a cyberattack from rogue OpenAI models. During the attack, Hugging Face turned to a self-hosted Chinese model that lacked the same guardrails as the U.S. models. This move comes as U.S. lawmakers consider measures to curb the adoption of Chinese AI models, which are often accused of extracting knowledge from U.S. systems through distillation. While many of the most capable open-source models are Chinese, tech giants have urged policymakers to avoid premature restrictions that could stifle competition or drive innovation overseas.