Nvidia, Palantir, Hugging Face join 34 others in race to defend open-weight AI from cyber threats

Thirty-seven companies and groups launched the Open Secure AI Alliance to strengthen security for open-weight AI models and their surrounding software.

Members include Nvidia, Microsoft, Cisco, IBM, Palantir, Hugging Face, and the Linux Foundation, but not closed-model labs OpenAI or Anthropic.

The alliance will build tools and methods to find vulnerabilities, patch them, and improve model provenance and deployment safety.

The move reflects growing concern that open-weight AI should be treated as a cybersecurity priority, not just an AI safety issue.