Clipnote Privacy Policy

Last updated: October 27, 2026

Summary

Clipnote does not collect, sell or share any personal data. AI summaries, transcription and notes are all processed on your own computer. Clipnote has no servers.

Where Clipnote runs

• YouTube: on youtube.com video pages the panel appears automatically.
• Other sites: Clipnote runs on a page only after you click its toolbar button on that tab. It does not run on, or read, any other page.

What Clipnote reads

• Captions the site already provides for the video you are watching, to build the summary and the searchable transcript.
• Playback time of that video, so notes and key moments can jump to the right place.
• Audio, only when you press “Listen & transcribe”: the sound of that video (or of that tab, if the site does not let Clipnote read the video directly) is turned into text by a speech recognition model running on your computer. The audio is never recorded to disk and never sent anywhere. Listening stops when you press Stop, close the panel or leave the page.

Video files on your computer

You can open a video or subtitle file from your computer in Clipnote’s own page. The file is played and processed only inside that page on your computer; it is never uploaded.

Offline mode and the network report

The panel shows how many requests Clipnote’s AI engine has made: model file downloads from Hugging Face, and requests anywhere else (always 0). After the models are downloaded you can turn on Offline mode, which blocks every request from the engine to the outside.

What Clipnote stores

Your notes, text transcribed with “Listen & transcribe”, AI summaries of videos you opened, the Library (title, address and captions or transcript of videos you summarized, so you can search them later), and a few settings (panel open or closed, spoken language) are saved with Chrome’s local extension storage on your computer. They are never uploaded. You can delete one video or clear the whole Library from the Library page; uninstalling Clipnote deletes everything.

Downloads Clipnote makes

• The first time you use “Listen & transcribe”, Clipnote downloads the open Whisper speech recognition model (about 80 MB) from Hugging Face (huggingface.co).
• The first time you use the AI summary, Clipnote downloads Google’s open Gemma 3 1B language model (about 760 MB) from Hugging Face. It runs on your own computer using your graphics card (WebGPU).
• Both models are kept in your browser’s cache. Only the model files are downloaded; no audio, captions, text or personal data is sent anywhere.

Gemma is provided under and subject to the Gemma Terms of Use found at ai.google.dev/gemma/terms.

What Clipnote does not do

• No accounts, sign-in, analytics, tracking or advertising.
• No browsing history is read.
• No audio or text leaves your computer.

Changes and contact

If this policy changes, the new version will be posted on this page with a new date. Questions: open an issue at github.com/sintori2/clipnote/issues.