The real ones get paid.

AI avatars are everywhere now. They sell skincare, front campaigns, read the news. Most of them were never anyone — and a startling number were someone, scraped and synthesized from a real person who was never asked, never credited, and never paid.

That second category is a problem the industry keeps pretending it doesn't have. A generated face that resembles a real person is, in most of the world, a right-of-publicity exposure waiting to be litigated. The likeness of a human being belongs to them. Borrowing it without consent is the oldest extraction there is, wearing new software.

So we built the opposite of a scraper. On Manekkin, every avatar begins with a real, verified person. They prove who they are once, with an ID check and a liveness selfie. They train an avatar on their own likeness. They write the terms — where it may appear, for how long, whether they keep final cut on every frame. Nothing is generated against a face that hasn't agreed to it.

And when a brand uses that avatar, the person behind it keeps the larger share — eighty-five percent of every deal, on a standing licence they can revoke whenever they choose, not a one-off buyout that signs away their face forever. The platform takes fifteen percent to keep the lights on and the verification honest. That's the whole arrangement.

Proof matters because trust does. Drop any Manekkin render into /verify and you can read, in plain words, whether a consenting person stands behind it, and the license that governs how it may be used. The credential travels with the image. The record is independent of us. Anyone can check it. That should be table stakes for synthetic media.