Privacy
6 min
read
On-Device AI vs Cloud AI in Notes Apps: What Actually Matters for Privacy
Most AI-powered notes apps send your writing to a server to power their features. Here's exactly what that means, what the alternative looks like, and why it matters more than the privacy policy claims.
When a notes app tells you it uses AI to summarise your notes or find connections between them, there's a question worth asking that most people don't: where does that AI run?
The answer determines who can read your notes, what they can do with them, and whether "private" means anything at all.
How cloud AI works in practice
Most AI features in consumer apps work the same way. You write a note. That note gets sent to a server — the company's own, or more likely a third-party AI provider's. The AI model processes it, generates a summary or a suggestion, and sends the result back to your device. Your note has now existed, however briefly, on a computer that isn't yours.
The privacy policies are usually careful. They say things like "we don't store your content" or "data is encrypted in transit." These are true statements that don't tell the whole story. Even if content isn't permanently stored, it was processed. It passed through infrastructure you don't control. The people running that infrastructure made decisions about how it was logged, how long it was retained before deletion, and what their own AI training pipelines exclude.
For a shopping list or a grocery budget, this is probably fine. For meeting notes, client conversations, personal journals, or early-stage ideas you haven't shared with anyone — it's worth thinking about more carefully.
What on-device AI means
On-device AI means the model runs on your hardware — your phone's processor, your laptop's chip. The note never leaves your device to power the AI feature. There's no server in the loop. The encryption question becomes moot because there's nothing being transmitted.
The tradeoff used to be significant: on-device models were much less capable than cloud models, and they drained battery faster. Both of those gaps have closed substantially. Modern on-device AI models — running on Apple Silicon and newer Android chips — are capable of high-quality summarisation, semantic search, and connection-finding at a quality level that would have required cloud infrastructure two years ago.
What to look for
When evaluating an AI-powered notes app, the questions worth asking are:
Does the AI feature require an internet connection to work? (If yes, it's almost certainly cloud-based)
What does the privacy policy say specifically about AI processing, not just data storage?
Can you use AI features offline?
The answers tell you more than any privacy badge or marketing claim.
The notes worth protecting aren't always the ones that feel sensitive. Sometimes the most private thing is an early idea that nobody else has heard yet.
