Nvidia's AI Security Alliance Shows Results After One Week

The Open Secure AI Alliance, a week-old industry group led by Nvidia with over 120 members, has already formed a working group called SAFE and is presenting open proposals on AI cybersecurity incident reporting and analysis. The Linux Foundation is managing the proposals, unveiled at the Black Hat conference in Las Vegas. Members are also pooling open source tools, including Nvidia's LLM scanner Garak, Amazon's Strands Agents, and Okta's agent identity tech. Notable absences include Anthropic, OpenAI, and Google, despite the latter two signing the original open letter that spawned the group.