Local Private LLM — Windows privacy

Local Private LLM runs AI inference on your Windows computer. It has no remote application backend, cloud AI account, API-key requirement or app telemetry. Model downloads are user-initiated online actions: model distribution services receive the IP address, requested model and connection metadata, and each model has its own license. Saved conversations are protected with Windows user-account encryption (DPAPI). Models and app data are stored under %LOCALAPPDATA%\LocalPrivateLLM. The app does not encrypt model files. Markdown exports are unencrypted files saved where the user chooses. Temporary chat is not written to conversation history, but data remains in process memory and may be retained by paging files, crash dumps, backups or administration tools. Local encryption does not protect against malware or another process with the same Windows-user privileges. The app starts a loopback-only model runtime and disables cloud inference. It shuts down the runtime it started on normal exit; an abrupt termination may prevent cleanup.