NVIDIA has established the Open Secure AI Alliance with 27 founding members including Microsoft, SpaceX, Dell, and the Linux Foundation to enhance cybersecurity defences. The coalition contends that effective cyber protection requires access to both proprietary and open-source AI systems, enabling security researchers to employ different tools depending on operational circumstances.

The alliance's rationale stems from a practical incident: when Hugging Face experienced a security attack, commercial closed-source models declined to assist with forensic analysis due to built-in safety guardrails, whereas open-weight systems successfully identified and contained the breach. Members are contributing distinct capabilities whilst simultaneously urging governments to classify open AI models as defensive assets rather than regulatory liabilities—a position that directly counters the Trump administration's reported consideration of restrictions on Chinese open-source AI systems.