Last updated 14 November 2026
Privacy Policy
Scribiz turns a video or an audio file into a transcript, the text and scenes shown on screen, a summary and chapters. This policy covers the website, the MCP server, the API and the command-line tool, and the Mac app once it is out. If something here is unclear, email [email protected].
On this page
Who we are
Scribiz is a service run from scribiz.com by Superleuk LLC, 500 Westover Dr #32164, Sanford, NC 27330, United States, and made by Ilias Ism (il.ly). In this policy, we means Superleuk LLC. Questions, privacy requests, copyright notices and abuse reports go to [email protected].
What we collect
- What you submit. Links, and audio from files you upload on the web or that the Mac app or the CLI sends for hosted processing. A video file is never uploaded whole. The Mac app and the CLI send audio, and for Watch a small low-resolution copy of the picture with no sound.
- What we make from it. The transcript, the on-screen text and scene notes, the summary and the chapters, and the private page that shows them. When you are signed in to a Scribiz account in the Mac app, the CLI or the local MCP server, the resulting transcript and summary (never the audio or video file) are also saved to your account so they appear in History on all your devices, kept for 30 days, and deletable at any time, even when the audio was processed locally or with your own key.
- Account details. If you sign in, your email address and name from Google or from the email you give us, your plan, your minute and caption credit balances and the API keys you create. Keys are stored as hashes, so we cannot read them back.
- Usage records. For each job: the mode, where it came from (web, Mac, CLI, MCP or API), how many minutes or caption credits it used and whether it worked. We use them for limits, billing and finding faults.
- Technical data. Your IP address and basic request details, used for rate limits and to stop abuse. For the limits that apply without an account, we turn your IP address into a code that changes every day, so the same person has a different code tomorrow and our usage records do not hold the address. Before Listen or Watch without an account we may ask Cloudflare Turnstile to check that you are not a script. Cloudflare runs that check under its own privacy terms.
What stays on your device
With the CLI and the Mac app, a local file is never uploaded whole. The audio is read on your computer. With a local speech model the audio never leaves your computer; with your own Gemini or OpenRouter key the audio goes from your computer to that provider and not through Scribiz. When you are signed out of Scribiz, the transcripts and summaries stay on your computer only (in ~/.scribiz/cache for the CLI, which --no-cache turns off, and in the app's folder on your Mac). When you are signed in to a Scribiz account, the resulting transcript and summary (never the audio or video file) are saved to your Scribiz account so they show in History on all your devices, kept for 30 days, and you can delete them at any time. Your API keys stay in ~/.scribiz/config.json (readable by you only) or the macOS Keychain.
On the website, your browser reads the audio out of the file you choose and uploads only that, and only after you press the button. If the browser cannot read the audio, the upload is refused. We never fall back to sending the video.
Analytics
This site uses first-party web analytics operated by Superleuk LLC on our own infrastructure, so visit data is not shared with third-party analytics companies. It records the pages you open, where you came from (including a ref or UTM tag on the link), your country, your device type, browser, screen width and operating system, how long a page stays open, how far you scroll, outbound link clicks, and a small set of named events, such as a job starting, finishing or failing, a format being downloaded, a click on the Mac download or a pricing button, a checkout starting or completing, an MCP snippet being copied, and a click on a button in a blog guide or its pop-up being shown. The events do not include the link you paste or the file you drop. The analytics script does not load when you open a result page (one that starts with /r/). If you reach a result from another page of this site without reloading, the script follows that move and can record the result's address, so do not treat a result link as secret from our own analytics.
Our analytics sets first-party cookies on scribiz.com. ml_vid and datafast_visitor_id recognise a returning browser and last 365 days. ml_sid and datafast_session_id group the pages of one visit and last 30 minutes. You can block or clear all of these in your browser settings, and the site works without them.
Cloudflare, which serves this site, may also load Cloudflare Insights (a short script from static.cloudflareinsights.com). It counts visits and how pages load. It sets no cookies of ours. Cloudflare handles that data under its own privacy terms.
What leaves your device, and where it goes
- YouTube caption retrieval. We send public video, playlist and channel identifiers to a caption retrieval service to obtain metadata and existing caption text. This route sends no video or audio to a model. The retrieval service receives those identifiers and our server request, not your account email or Scribiz key.
- Links and audio sent for hosted processing go to our servers, and from there to the speech model named in the next paragraph for transcription and to the Gemini API from Google for analysis. The dedicated YouTube transcript and bulk caption routes retrieve existing text only. When you explicitly choose media processing for a public YouTube link, Auto mode uses its captions first. YouTube blocks the download from our server, so to listen we pass the link to Google, whose model reads the public video. We send Google the link, and we make no copy of the video. For other sites and direct media links, our server downloads the audio (and for Watch a low-resolution copy of the picture), sends it to those models and deletes it when the job ends.
- Audio that our servers transcribe (a file you upload, a link whose audio our server downloads, a voice note or file you send in Telegram or WhatsApp) is transcribed by a speech model from ElevenLabs, which our server reaches through OpenRouter, or by Gemini when that model is not available. The audio travels inside the request and we store no file with either company. We ask OpenRouter to use only providers that do not keep what they are sent or train on it. That is their commitment, under their own terms, and it can change. Summaries, proofreading and what is on screen are done by Gemini. A YouTube link that Google reads itself, and audio you send with your own Gemini key, do not go to ElevenLabs or OpenRouter.
- In the command-line tool and the Mac app you can choose the speech model. With your own OpenRouter key the audio goes from your computer to OpenRouter, which sends it to the speech provider, under your key’s terms and OpenRouter’s. With a model that runs on your computer, the audio is read on your computer and never leaves it. Your OpenRouter key is stored in your config file or your Keychain and goes only to OpenRouter. When you are signed in to a Scribiz account, the resulting transcript and summary are saved to your Scribiz account so they show in History on all your devices (kept 30 days, deletable at any time); when you are signed out, they stay on your computer only.
- What Google may do with what we send it depends on its terms for the kind of key used: its terms for paid use say it does not use prompts and responses to improve its products, and its terms for free-tier use say it may. Those are Google’s terms and they can change, so read the current Gemini API terms if this matters to you.
- The result page for a YouTube video shows it in an embedded YouTube player. Opening the page loads Google’s player script and the player, and Google then receives your IP address and the video’s id, under Google’s own privacy policy.
- When you sign in to Scribiz from the CLI (scribiz login) or the Mac app, hosted runs send audio from your computer to our servers, and from there to the speech model or the Gemini API from Google, as for hosted processing. With your own Gemini key, in the CLI, the local MCP server or the Mac app, audio goes from your computer straight to Google under your key’s terms. For a public YouTube link that the CLI reads through Gemini, only the link goes to Google, and Google reads the video from YouTube. When you are signed in to a Scribiz account, the resulting transcript and summary are saved to your Scribiz account so they show in History on all your devices (kept 30 days, deletable at any time); when you are signed out, they stay on your computer only. A free-tier key may let Google use your content to improve its products. Turn on billing for the key to opt out.
- Files we upload to Google for a job are deleted when the job ends. If a delete ever fails, Google removes uploaded files itself within 48 hours.
- We use a small number of service providers for hosting, sending emails (sign-in links, payment receipts and partner notices), error reports (when something fails on our servers, a short record goes to Sentry: the error and the id of the job or account it happened to, never your files or results) and bot protection. They handle data only to do that job. We do not sell your data, and Scribiz does not use your videos, audio or transcripts to train models of its own.
How long we keep things
- Media. Uploaded audio and any temporary copy are deleted when the job ends.
- Results. Without an account, a result is a private link that is kept for 24 hours and then deleted. A result kept by an account lasts 30 days, unless you delete it sooner.
- Cache. To avoid doing the same work twice we keep a cache, keyed by the video’s identifier on the site it came from. It holds text, never media, and only for videos of a platform such as YouTube, never an upload or a direct media link. Video details are kept for 24 hours. Successful raw YouTube caption snapshots are kept for up to 365 days in a separate cache capped at 8 GiB. Other captions are kept for up to 30 days and generated text (transcript, scene notes, summary and chapters) for up to 90 days. A cache entry may be removed earlier when its storage budget is reached. It is used only when someone asks for that same video. It is never shown as a public page, and we purge it when a takedown notice requires.
- Account and usage records. Until you delete your account. Billing records, once payments exist, for as long as the law requires.
- Server logs. Kept briefly, and used only for security and debugging.
Results are private
A result lives at an address with a long random identifier. Search engines are told not to index it, and we never publish a result as a public page. Anyone who has the link can open it until it expires or you delete it, so sharing the link shares the transcript. Dedicated bulk caption batches and their exports require the account or API key that created them. They are kept for 30 days. We do not publish results in a public directory.
Partner links
A partner is someone who shares a Scribiz link that ends in ?via= and a name. When you open one, we record the visit for that partner: the time, your country as our network provider reports it, the website you came from and whether you are on a phone. We do not store your IP address, and the visit is not tied to you. Your browser keeps the partner's name and the time for 30 days.
If you sign in within those 30 days and have not paid us before, your account is linked to that partner once. Payments you make in the next 12 months then earn the partner a share. The partner sees that a customer paid, what was bought and when, but never your name, your email or anything you used Scribiz for. Partners who apply give us their email, their website, who they share with and where to send their payouts, and we keep their clicks, referrals and payouts for as long as they are a partner and as long as the law requires for payments.
Payments
Payments are taken by Stripe, on Stripe's own page. Stripe handles card details, and we do not see or store card numbers. We keep your plan and the status of each purchase. Stripe keeps its own payment records, also after you delete your account, under its own privacy policy. We keep a record of each payment (what was bought, the amount and the date) without your name or email, so that refunds and our accounts stay right.
Your choices and rights
You can delete a result from its page whenever you like, and ask us to delete your account and everything tied to it. You can ask for a copy of your data, ask us to correct it, or object to how we use it. Where the law gives you more, for example in the EU, the UK or California, you have those rights too, including the right to complain to your data protection authority.
Email [email protected]. We will answer within 30 days. To report a copyright problem, see the takedown page.
Children
Scribiz is not meant for children under 13, and we do not knowingly collect their data.
Security
We use HTTPS, store API keys as hashes and limit who can reach our systems. No service is perfectly secure, so do not submit anything you could not bear to see leak.
Where data is processed
Our service providers and Google may process data in countries other than yours. Where the law requires a transfer mechanism, we rely on one.
Changes
When we change this policy we change the date at the top. If a change affects what we do with your data in a way you would not expect, we will say so on the site before it takes effect.