NVIDIA and 121 firms launch Open Secure AI Alliance

NVIDIA and 121 companies have launched the Open Secure AI Alliance to develop and share open technologies, tools, and techniques aimed at safeguarding software and AI agents. Building on the Linux Foundation’s Akrites initiative and OpenSSF community work, the alliance focuses on remediating and disclosing vulnerabilities through open-source methods.

NVIDIA advocates for open models and harnesses to democratize defensive capabilities and increase transparency. The company noted that closed AI tools can sometimes hinder forensic analysis by failing to distinguish between attackers and defenders. By utilizing open frontier defensive tools, critical industries can build security systems across multi-vendor ecosystems to avoid single points of failure.

In conjunction with these efforts, NVIDIA has released an analysis regarding security within the AI agent stack. As enterprises adopt more autonomous, long-horizon agents, the company emphasizes the need for layered security systems. This includes defining clear boundaries across the model, the harness, and the runtime layers to manage risks such as unauthorized internet access or unapproved interactions with external systems.