Why we'll never use your photos to train AI

Almost every photo service today promises your data is 'safe'. But 'safe' means little once you open the terms. They often contain a line granting a 'worldwide, royalty-free licence' to use your content — among other things, to 'improve' products. That last word is exactly where models get trained.

What 'no training' really means

At Ohhi the principle is simple: your photos are yours. We get no licence to reuse them, analyse them for ads, or feed them to a machine-learning model. Not by us, and not by a third party.

That sounds obvious, but it's an active choice. It means features that can only exist by training on your images — automatic face recognition across your whole library, say — we deliberately build differently. Recognition is opt-in and stays on our own servers in Europe.

A promise you can't enforce isn't a promise. It's marketing.

How we enforce it

We make it concrete in three ways:

• Contractually — our terms grant us no licence to use your photos for training. There's no hidden clause.
• Technically — images sit on servers in Europe, separated from any analysis or model process.
• Provably — we publish what we do and don't do with data, in plain language, and keep it current.

The result is a service that's more boring than many rivals — no clever feed that gets to know you, no surprising 'memories' popping up from nowhere. That's exactly why we believe calm beats an algorithm. Want to know where your photos actually live? We wrote it up in servers in Europe: what it really means. Ready to start? See the plans — there's always a free tier.