Where the data comes from.
Most AI is trained on whatever could be scraped. Real Cut is trained on three sources, all accounted for: edits we own, footage we license, and the projects people choose to contribute — for a paid human edit in return. Nothing scraped. Every editor who teaches the model is paid and credited. This page keeps the count.
TRAINING-CIRCLE PROJECTS
0
PEOPLE WHO OPTED IN, FOR A PAID HUMAN EDIT
CONSENTED INTERACTION DATA
0
ACCOUNTS SHARING EDIT DECISIONS — NEVER FOOTAGE
EDITORS ONBOARDED
9
PAID PER SESSION · CREDITED BY NAME
FOOTAGE SCRAPED
0
AND IT STAYS THERE
01The three sources.
Owned. Real edits we made ourselves, with the raw footage, the finished cut, and the decisions in between. Licensed. Footage we pay for, cleared for training. The Training Circle. Projects users opt in — their raw footage plus both edits (the AI's and a paid editor's) plus the decisions. Used only to train and evaluate the model. Never sold, never used for ads, never used to identify anyone in the footage.
02How the Training Circle works.
You are asked once, per project, only after you have a finished cut in hand — never at signup, never buried in the terms, never in the way of your upload. If you opt in, a professional editor we pay also cuts your footage and you keep that edit for free. You can leave any project, or leave entirely, from your account — we stop all future use and delete your materials. The one thing we can't do, and say so plainly: un-teach a model version that already learned. Your footage simply never appears in any future training run.
03The editors, credited.
Every editor who teaches RCM 1 is paid per session and credited by name unless they ask otherwise. As real sessions run, their credits appear here — the same promise the footer makes, kept in public. Editors apply here.