Nvidia and a group of dozens of technology companies launch the Open Secure AI Alliance, a new effort to develop and share open-source tools for AI cybersecurity defense. Multiple outlets report that founding members include Nvidia along with companies such as Microsoft, SpaceX, IBM, and Palantir, among others. The alliance’s stated goal is to help defenders better test, audit, and protect AI models and AI-based agents by building software that can identify, patch, and disclose security vulnerabilities.
Several sources link the initiative to recent concerns and incidents involving frontier AI systems, including reporting about fallout following a cyber incident involving OpenAI’s systems during an internal security evaluation. At launch, Open Secure AI Alliance members explicitly exclude some major AI companies: the coalition does not include OpenAI, Google, or Anthropic, according to reporting from The Verge, CoinDesk, and other outlets. Some coverage also notes the effort builds on or aligns with work already underway in open-source security initiatives.
Overall, the alliance focuses on making AI security capabilities more broadly available through shared open tooling for infrastructure defense and vulnerability management.
