More than 50 technology companies and organizations joined NVIDIA to launch the Open Secure AI Alliance, a coalition dedicated to building shared cybersecurity tools and frameworks for the age of AI agents.
Inaugural partners span cloud computing, enterprise software, cybersecurity and AI research, including Microsoft, CrowdStrike, Hugging Face, IBM, Palo Alto Networks, Databricks and SpaceXAI. The alliance builds on existing Linux Foundation initiatives and aims to develop open-source technologies for defending critical infrastructure against AI-enabled threats.
NVIDIA contributed its Labs Object-Oriented Agent research framework, which helps make agent behavior easier to test, trace and audit. Other members contributed tools for identity verification, safe model storage formats and multi-model security scanning — Microsoft’s MDASH harness, for example, orchestrates specialized AI agents to discover and prove exploitable bugs.
The coalition argues that open models are critical for cybersecurity because they let defenders inspect, adapt and run defensive tools on their own hardware. Making security technology open and auditable gives more defenders the ability to strengthen systems, the group contends, rather than relying on a small number of closed providers. The alliance also called on policymakers to treat open AI security tools as defensive assets rather than liabilities in regulation.
The launch comes weeks after the Hugging Face security incident, where closed AI tools blocked essential forensic analysis during a breach response.
