Why we'll never use your photos to train AI

Almost every photo service today calls your data 'safe'. That word holds up poorly the moment you actually open the terms. They often state that you grant the service a 'worldwide, royalty-free licence' to your content, among other things to 'improve' products. In practice, training models tends to fall under exactly that word, 'improve'.

What 'no training' means for us

Our starting point is that your photos are yours and stay that way. We ask for no licence to reuse them, we do not analyse them for ads, and we do not feed them to a machine-learning model. That applies to us and to the parties we work with.

That choice comes at a price. A feature like automatic face recognition normally only works once you train on someone's entire library, so we build it differently. At Ohhi you turn recognition on yourself if you want it, and the computing happens on our own servers in Europe.

A promise about your data is only worth something if you can verify it. That is why we put it on record not only in words, but also in the technology.

How we put it on record

We do not leave it at that promise. We make it concrete in three places:

• In the terms. They state in black and white that we get no licence to use your photos for training. There is no clause at the bottom that quietly reverses it.
• In the technology. Your images sit on servers in Europe, separated from anything to do with analysis or models.
• In what we publish. We describe in plain language what we do and do not do with your data, and we update it as soon as something changes.

The result is that Ohhi feels more sober than many competitors. There is no clever feed that gets to know you, and there are no 'memories' that suddenly pop up. That is a deliberate choice, and the same reason we think calm works better than an algorithm. Where your photos do live, you can read in servers in Europe: what it really means. If you want to see it for yourself, you can start for free via the plans.