The fine print, in plain language
Privacy policy, terms of use, and where the numbers come from
Last updated: October 19, 2026. Written to be read, not skimmed past. If anything here is unclear, ask us before you rely on it.
Privacy policy
Your video never leaves your computer. Rundown Clipper opens the newscast file straight from your own disk, inside your own browser. There is no upload of the video, no copy on a server, and no account, library, or history where it could sit.
What does get sent: short pieces of the show's audio — no picture — for transcription. The resulting transcript and any optional newsroom script you paste or import are sent to AI for story matching, titles, and checklist suggestions. Only include script material your station permits you to share. The app does not save scripts to an account or library; closing the tab clears the local script.
No sign-up, no tracking. We don't ask for your name, email, or payment details to use the tool, and we don't build profiles about you or your station from your newscasts.
The clips you make are rendered on your own computer too. They are downloaded by you directly and are never stored by us.
If that ever changes — for example if a future version offers cloud processing for very large files — this page will be updated first, and it will say plainly what is uploaded, where it goes, and when it is deleted.
Terms of use
You must have the rights. Only load and clip video you are allowed to use — your station's own broadcast, or footage you have permission from. Using someone else's footage without permission can be copyright infringement, and that responsibility stays with you.
It's an assist tool, and a person makes the final call. Story boundaries, suggested titles and descriptions, start and end times, post scores, and brand-safety ratings all come from a checklist of rules applied the same way to every story — and the checklist will get some wrong. You agree to review every clip — watch it, check the times, read the wording — before anything goes to air or out on social.
The numbers are estimates, not promises. Scores and safety ratings are the AI's opinion based on how stories generally perform online. They are not guarantees of views, revenue, or advertiser acceptance, and they are not yet based on your channel's real performance.
No warranty. The tool is provided as is. We aim for accuracy and availability, but we can't promise the transcription, the story finding, or the exports will always be perfect or uninterrupted — so build in a human check before anything is published.
Fair use. Don't use the tool to produce misleading or fabricated content, to clip material you don't have rights to, or to attempt to break it.
How we get this information
Every number the report card shows you is produced the same way, from three sources you can check yourself:
1. What the AI actually reads. The spoken words of your show, turned into text, plus any newsroom script you choose to attach. It never sees the picture or pulls in outside performance data. Scripts help match actual story slugs and suggest leaving out separate anchor leads when the package start is clear. Planned script times are not treated as recording times; unmatched or unaired items are not proof of a clip. Start and end suggestions still need your review.
2. A real, measured test run. During development, a full 49-minute local newscast was processed end to end: 42 stories found, each with a title, description, post score, and safety rating, and finished clips exported in both wide and vertical formats. The AI usage — and the cost per show — was measured in that run, not estimated.
3. Published benchmarks. Processing costs come from the AI providers' public list prices (roughly $0.05–$0.10 to analyze a one-hour newscast, because only the audio is analyzed, whatever the video size). Competitor pricing comes from the published rates of enterprise clip suites, which run $50,000–$250,000 per year and are aimed at sports leagues and national networks, not a single station's producer.
What this does and doesn't back up. It backs up the cost figures and the workflow claims. It does not yet back up the view or revenue predictions — those become verifiable the moment a beta station's real clips can be compared against its real YouTube and Facebook numbers, which is exactly what the beta is for. Until then, treat every score as a starting shortlist, and your judgment as the final word.
What's AI, what's not
Two steps use AI, and everything else is ordinary browser code. When other pages on this site say "the checklist," that checklist is an AI model reading your transcript — so here it is spelled out plainly, before anyone has to ask.
AI step 1 — transcription. Short pieces of your show's audio (never the picture) are sent to a speech-to-text model and turned into words. Nothing is stored once the job is done.
AI step 2 — the story checklist. An AI model reads the transcript and any attached script, suggests where each story starts and ends, matches script slugs, writes the titles, descriptions, and social captions, and produces the post scores, safety labels, and platform suggestions. Leaving out an anchor lead is a suggested start, not a frame-accurate cut. There is no direct ENPS or iNEWS connection. The same rules are applied to every story, but it is an AI model doing the reading — and it will get things wrong. That is why every clip still waits for your approval.
Not AI — anything you can see or touch. The start and end buttons, the timeline, the wide or vertical choice, the full-screen vertical crop and its left/right slider, caption editing, and the rendering of the downloadable clip are all ordinary code running on your own computer. No AI picks your clips, and nothing exports without your sign-off.
Notes for beta testers
A few things worth knowing before your first run. None of these are bugs you did anything wrong on — they're just how the tool works today.
If the site won't load on a station computer, it's the network filter, not the app. Some station networks (Fortinet and similar security filters) block the site by mistake. Try a phone or a personal device — if it loads there, ask your IT team to allow rundownclipper.com on the station network.
Use a modern browser, ideally Chrome. The clipping engine runs inside the browser itself, so older browsers may struggle or refuse.
Stay on the tab while a clip is being made. The export pauses if you switch away and finishes when you come back — no jumps, no corruption, it just waits. There's also a reminder under the download buttons.
File sizes: up to about 2 GB per file. That covers most recorded newscasts. Very large files can be heavy on the browser — if a big one crashes the tab, try a shorter recording and tell us; it helps us plan the next version.
Try your non-English shows too. Story finding, titles, and captions work in many languages (use the language picker under the upload box). Spanish and German are strong; languages like Farsi and Kurdish may have more mistakes on names and places — please review those closely and tell us what you find.
What we'd love feedback on: Did the story boundaries land where you'd cut? Were the titles and descriptions usable as-is? Did the post scores and ad-safety labels match your instincts? Any clip that came out jumpy, mistimed, or wrong — which story, which format.
Questions about any of this? Ask before you publish — we'd rather over-explain than have you guess.