What the AI does
You upload a take and say how you want it. FrameCutter writes down every word you said, ums included. The AI reads that transcript and your request, and decides the edit: which lines to keep, which take of a repeated line is best, where pauses and filler words come out, and where captions go.
The AI never touches the pixels. It writes the edit as a list of cuts and settings; regular video software then cuts your own footage to that list. Nothing in your video is generated: no faces, voices or footage that you didn't film.
What the AI sees
AI providers receive the text of your transcript and your requests, and a few still frames from your clips for some checks (for example, whether you're looking at the camera, and what's in a b-roll clip). They never receive your full video files, your email address or your payment details. A transcription provider receives the sound of your clips to turn speech into text.
Training
We don't use your footage to train AI. Our AI and transcription providers don't use your data to train their models. Our AI provider may keep what it receives for up to 30 days to detect abuse, and longer only if it is flagged or the law requires it. If you turn on “Help improve FrameCutter's editing” (off by default), your corrections and edits, never your footage, may be used to improve the editor for everyone.
How long your footage is kept
Full-quality originals are deleted 3 days after upload (14 on Pro); a lighter copy, the transcript and your edited videos stay while your account is active. Where your data is processed, and under which safeguards, is in the privacy policy.
Who sees your videos
Your videos are private to your account. Nobody else sees them unless you share a file or send us a problem report with a video attached.
All the details are in the privacy policy.