Last updated: 2026-09-01
Short version: we do not train AI on your models, and your 3D files are never sent to any AI system. Ever.
Here is exactly what happens.
When you list a model for sale, Minarium reads the preview images and renders you upload. These are the same pictures your customers see on your store page, and the same ones you already post publicly anywhere else you sell.
Those images get converted into a list of numbers describing what's in them. That's what lets search understand that someone typing "armored dwarf with an axe" wants your armored dwarf with an axe, instead of only matching whatever you happened to name the file.
That's the whole thing. It's search.
Vault search works on file names and metadata only. Nothing from inside a Vault archive is ever sent to an AI system: not the files, not the images inside them, not the list of what they contain.
Our own systems do open your archives, on our servers, for two mechanical jobs: reading the list of files inside so you can browse a collection, and pulling out one image to use as the thumbnail in your library. No AI is involved in either, and nothing from that process leaves Minarium.
Your Vault is private and stays that way. Sharing features are not offered and are not planned.
Marketplace image processing runs through a third-party vision service. Their terms exclude data sent through their API from being used to train their models, and we have not opted in to any data sharing. They keep request data for a short window for abuse monitoring, then delete it. That's standard for services like this.
Nobody could rebuild one of your sculpts from anything in that pipeline. There is no mesh in it.
This is written into our Terms of Service at Sections 16.8 and 18, so it's a commitment and not a blog post. If you've heard otherwise, or have a question this doesn't answer, email [email protected] and we'll answer directly.