Foundry Local is Microsoft’s local AI deployment layer for running model inference on customer-controlled infrastructure instead of only in Microsoft-managed cloud services. It is designed for on-premises, edge, or otherwise controlled environments where organizations want the model to stay closer to their own systems, data, and policies.
In cyber security, this matters because local AI changes the trust boundary. It can reduce exposure of sensitive prompts and outputs to external services, but it also expands the assets defenders must secure: model files, API endpoints, update channels, identity controls, logs, and the host platform itself. Attackers may target these layers to steal models, poison outputs, or abuse local inference services for data access. Defenders use it to support data residency, offline operation, and tighter governance, but they still need patching, monitoring, certificate management, and access control.