Content and Licence

Your prompts, your uploaded images and video, your LoRA files and your flows remain yours. We do not become an owner of them by you putting them in the studio.

You give us permission to store and process that material only to run the service: keeping it in your library, showing it back to you, and sending what a generation needs to the provider that performs it. Nothing beyond that.

We do not use your prompts, uploads or results to train, fine-tune or improve any model, and we do not sell them or hand them to anyone for that purpose. We are not in the business of building models; we buy access to models other people built.

For the AI assistant, which runs on a language model reached through OpenRouter, every request we send carries two instructions: route only to providers that do not collect data, and only to endpoints that retain nothing. That is set in our code on every call, not in a dashboard setting that somebody could quietly flip.

What we can promise stops at the providers we choose and the settings we set. We cannot audit what a provider does inside its own systems. If that difference matters for your work, treat any AI service the same way and do not send material you cannot afford to have leave your building.

We claim no ownership of your results, and we place no restriction of our own on using them commercially. Every one of the 46 models in the studio is marked by its provider as licensed for commercial use , we checked all of them, not a sample.

You can therefore use what you generate here in client work, in advertising, in products, and for anything else that is lawful, subject to the Acceptable Use policy.

This is where most services stay vague, so we will not. Under Dutch and European copyright law, protection requires a human creator. Material that is generated purely by a model from a prompt may well not be protected by copyright at all , not for us, and not for you.

In practice this means two things. You are free to use the result. But you may not be able to stop somebody else from using something very similar, and you cannot assume you hold an exclusive right in it. The more of your own creative choices go into the work , your own photograph as the source, your own style model, your own editing , the stronger your position becomes.

We cannot give you legal advice about your specific case, and nobody can promise you copyright in a generated image. Anyone who does is selling you certainty they do not have.

You warrant that you are allowed to use what you put into the studio. Concretely:

• Uploads. The photos and video you upload are yours, or you have permission to use them this way.
• LoRA files. You may use the style model you bring, and it was trained on material you were allowed to train on. A LoRA of somebody’s face or of a living artist’s work is your responsibility, not ours , and in the case of a real person, generally requires their permission.
• Faces and names. Generating a recognisable real person carries portrait rights and personality rights on top of copyright, and those belong to that person.
• What you publish. Checking the result before you put it in front of a client or the public is your job. Models get things wrong, including things that look right.

If a third party holds us liable because of what you made or uploaded, you cover that claim.

Under the EU AI Act, providers and users of AI systems have transparency duties, and for synthetic images, audio and video that increasingly means labelling them as such , particularly deepfakes of real people, and material published to inform the public.

The studio does not label your output for you and does not embed a watermark. Whether you have to disclose, and how, depends on what you publish and where. It is worth checking; it is not something we can decide on your behalf.

We do not use your results to promote the studio. If we ever want to, we will ask you first and you can say no without anything changing for you.

The starting flows, the model descriptions, the prompt guidance and the studio’s own software stay ours. You may use the templates freely for your own work, including commercial work; you may not repackage them or the studio as a competing product.

What you may not do with generated output is also limited by our providers: their terms forbid using output to train or improve a model that competes with theirs. That restriction sits in the Acceptable Use policy as well.