Hal Universal does not collect, store, transmit, or share any personal data of any kind.
It has no accounts, no analytics, no telemetry, no advertising identifiers, no third-party SDKs. There are no servers operated by the developer. All conversations, memories, and reflections are stored exclusively on your device in a local SQLite database, protected by iOS Data Protection — the file-level encryption iOS applies to app data when your device is locked.
When you use a local MLX model (Gemma, Llama, Qwen, or Dolphin), inference runs entirely on your device with no network calls of any kind. When you use Apple Intelligence, inference occurs on-device when offline. When connected, Apple's Private Cloud Compute may be used — your prompts are encrypted in transit and processed only in short-lived, non-persistent memory on Apple-controlled servers. Apple states no data is retained after processing. The developer of Hal Universal does not see or have access to any data Apple Intelligence processes.
MLX model downloads are fetched from Hugging Face over standard HTTPS at your explicit request. Once downloaded, models run entirely on your device.
You can delete all of Hal's local data at any time via Settings → Power User → Database → Nuclear Reset.
Questions: Support