CoreWeave Launches the CoreWeave Zero Trust AI Framework
CoreWeave, Wednesday, October 21st, 2026
CoreWeave unveils its Zero Trust AI Framework, letting customers bring their own security systems to CoreWeave Cloud.
CoreWeave announced the CoreWeave Zero Trust AI Framework at NVIDIA GTC Berlin, a customer-controlled security model that gives customers authority over access to their models, data and runtime environments.
The company says the framework lets customers bring the security systems they already run to CoreWeave Cloud instead of rebuilding around a provider-defined model.
CoreWeave argues that agentic AI moves sensitive data through prompts, embeddings, weights, traces and outputs, so traditional cloud trust models, where providers hold keys or privileged access, no longer fit.
The framework is aimed at enterprises running AI workloads that need tighter control over who can reach their assets.