On-Premise vs Cloud
Most AI scribes run in the vendor's cloud. Some enterprise deployments offer a virtual private cloud or on-premise option so PHI never leaves the health system's boundary. Ask vendors specifically whether inference calls out to a third-party LLM API.