Every notes app markets itself as private. Very few are. After building an AI notebook that runs entirely on-device, we've developed a short audit we run on any app that asks for our words. Run it on whatever you use today.
Question 1: Where does the text go when you type it?
This is the question that eliminates most apps immediately. If your notes sync across devices, they transit a server. That server might be well-run, encrypted at rest, run by a company with a good track record. But it exists, which means:
- It can be breached (every large sync provider has had incidents).
- It can be subpoenaed.
- Its operator can change the privacy policy after you've accumulated years of notes in it.
- Its employees can, in principle, access your content, whatever the policy says.
Sync is a legitimate feature with a real privacy cost. The audit question isn't "does it sync?" but "is the sync end-to-end encrypted, with keys only I hold?" If the answer isn't a clear yes, your notes are readable by the provider.
Question 2: What does the app know about you?
Open the privacy nutrition label or the app's analytics disclosure. Questions to ask:
- Does it use third-party analytics SDKs? (Most do. Every tap becomes a data point in someone else's warehouse.)
- Does it have an account system? Accounts mean an identity tied to your content on a server.
- Does the crash-reporting tool attach note content? Some do, in stack traces and attached logs.
The strongest answer, and the one we implemented: no analytics, no accounts, no third-party SDKs at all. An app that collects nothing has nothing to breach.
Question 3: What about the AI features?
This is the new frontier, and it's worse than most users realize. If your notes app has an "AI summarize" button, your notes are almost certainly being sent to a cloud API. Often with a different privacy policy than the notes app itself. Some providers use API inputs for training by default unless the app developer opts out. Almost no app surfaces this to you clearly.
The questions: Which model provider? What does that provider's API policy say about training and retention? Did the app developer opt out? If the app can't answer these, the answer is: your notes went to a server you didn't choose.
Our answer with the Offline AI Notebook: the models live on your phone. The question doesn't arise because no text ever leaves the device to be summarized.
Question 4: Can you verify any of this?
Trust is a process, not a policy page. Signals that an app is actually private:
- A privacy policy that's short and specific, full of "we don't" rather than "we may."
- Network monitoring. On iOS, an app like Privacy Insights or a DNS-level filter can show you what an app actually talks to. A notes app that talks to nothing is a notes app that can't leak.
- Offline behavior. Try airplane mode. If every feature works, the app probably has no server dependency for your content.
The uncomfortable summary
The default architecture of modern software is: your data, our servers. That architecture is convenient, and it is also the single largest source of privacy failures in the industry. Switching to apps where the data never leaves your device changes the question from "do I trust this company?" to "do I trust my own phone?"
We'd argue that's the only version of the question worth asking.