Nasu - Privacy Policy
Last Updated: July 24, 2026
Nasu is a privacy-focused audio transcription app that processes everything on your device. We are committed to protecting your privacy and to being transparent about exactly what the app does with your data.
The short version
- All AI processing — transcription, speaker identification, voice profiles, summaries, and titles — happens on your device. We do not operate any servers that receive your audio, transcripts, or voice data.
- We have no user accounts and no access to your recordings or transcripts.
- iCloud sync is off by default. If you turn it on, your data is stored in your personal iCloud account and handled by Apple (details below).
On-device processing
Your audio recordings, transcriptions, speaker labels, voice profiles, and summaries:
- Are processed exclusively on your device by locally-running models
- Are never uploaded to our servers or to any third-party AI service
- Are not used to train any models
Summaries, titles, and tag suggestions on supported devices use Apple Intelligence (Apple's on-device Foundation Models). This also runs on your device as part of the operating system.
Voice profiles (speaker recognition)
If you choose to save a voice profile so Nasu can recognize a person in future recordings:
- A profile is only saved after an explicit consent step — the app requires you to confirm the person has been informed and agrees
- Profiles are numeric voice signatures, not audio recordings, and are encrypted with a key stored in your iCloud Keychain — end-to-end encrypted, so neither we nor Apple can read them
- You can delete a person at any time; deletion removes their profile and every stored voice sample
- Speaker labeling works without saving any voice profile — profiles are strictly optional
Network access and model downloads
Nasu downloads AI models (speech recognition and speaker identification) when you first use the corresponding feature. Models are fetched from Hugging Face's content delivery network as one-way downloads — no user data, audio, or text is ever sent. After download, the models are stored locally and the features work completely offline.
iCloud sync (optional, off by default)
By default, everything is stored only on your device. If you enable Settings → iCloud Sync:
- Your recordings and transcripts are stored in the private database of your personal iCloud account, so they can sync across your devices
- That data is then handled by Apple on Apple's servers, under Apple's iCloud terms and Apple's Privacy Policy — Apple encrypts it in transit and at rest
- Voice profiles remain end-to-end encrypted with a key that only your devices hold, even when synced
- We (Zebel) never gain access to any of it — synced or not
- You can turn sync off at any time
Analytics
The app includes optional, anonymous usage analytics (Mixpanel) to help us understand which features are used. Analytics:
- Never include your audio, transcripts, speaker names, or any content
- Only cover feature-level events (e.g., "a transcription was completed", model choice, durations)
- Can be turned off entirely with one switch in Settings
- Are never sold or shared with third parties
For more information on Mixpanel's practices, see Mixpanel's Privacy Details.
Data storage and deletion
Your data is stored on your device (and, only if you opted in, in your personal iCloud account). Deleting a recording, a transcript, or a person's voice profile removes it from your storage; deleting the app removes all locally stored data. We hold no copies.
Changes to this policy
We may update this privacy policy from time to time. Any changes will be posted on this page with a revised "Last Updated" date.
Contact us
If you have any questions about this privacy policy or our privacy practices, please contact us at nasu [at] zebel.ai.