This policy covers everything called Samplecut: the website at samplecut.app, the API behind it, the Samplecut browser extension for Chrome and Edge, the Samplecut bot on Telegram, and the Samplecut app in ChatGPT, Claude and other chat apps. It describes what we actually collect, why, and what you can do about it. Samplecut does not sell personal data and does not show ads.
Using Samplecut without an account
Extracting, cutting and downloading audio needs no account. When you paste a link, the link is sent to our API, which fetches the audio, prepares it for the editor and keeps the working files for about an hour before deleting them. We keep a record of each video that has been extracted — its platform, ID, title, length, a small waveform preview and, where available, the start of its captions — so it can have a public page on the site. That record is about the video, not about you: it is not tied to your account, IP address or analytics ID.
Analytics and session recordings
We use PostHog (hosted in the EU) to understand how Samplecut is used and where it breaks. On the website this includes:
- page views and product events, such as a link being extracted or a cut being downloaded, with details like the video's platform, the format chosen and how long a step took;
- errors that happen in your browser;
- session recordings, which replay how the page changed and where you clicked, so we can see what went wrong in a session. A recording can include what was on the page, such as the link you pasted;
- a random ID stored in your browser so events from one visit can be grouped together, and your approximate location derived from your IP address.
The website passes that random ID to our API with each request, so a failed download can be matched to the session it happened in. Analytics is only switched on for the production site.
If you are in the EU, the EEA, the UK or Switzerland
We ask first. Until you accept, nothing is stored in your browser and there are no session recordings: page views and events are counted anonymously, grouped only by a code PostHog works out on its servers that changes every day, and nothing is passed to our API to link your requests together. If you accept, the random ID and session recordings described above are switched on. If you decline, anonymous counting continues. Whether you are asked is decided from your device's time zone. You can change your choice at any time from "Analytics settings" at the bottom of the home page. Your choice itself is remembered in your browser so we don't ask again.
Wherever you are, you can also block analytics with any common content blocker; the site keeps working.
Signing in with Google
You only need an account to publish cuts to the shared library. Sign-in is through Google; we never see your Google password. From Google we receive your name, email address and profile picture, and we store them with the username you choose and a session cookie that keeps you signed in. Your username is shown publicly next to cuts you publish. Once you are signed in, your analytics ID is linked to your account so that your events carry your username.
Published cuts
When you publish a cut, its source video, start and end times, label, waveform preview and your username become public in the library and on their own page, and may be shown to search engines.
Server logs
Our API keeps operational logs (requests, errors, timings) so we can keep it running. IP addresses are never written to these logs in readable form — only as a short one-way hash, which lets us tell whether repeated failures come from the same place without recording where that is. IP addresses are also used, in memory only, to rate-limit abuse. Logs are sent to PostHog.
The browser extension
The Samplecut browser extension adds download buttons to YouTube video pages. It works like this:
- Nothing is sent until you click. The extension reads the page you are on to place its buttons, but it only sends the video's URL to our API when you choose to download or open it in the editor. The extension does not read or send your browsing history.
- Analytics events. By default the extension sends a small set of usage events to PostHog: it was installed or removed, its buttons were shown or could not be placed, a download was started and whether it worked, a part of a video was marked, and the editor was opened. These carry the video's URL or ID where relevant, and a random ID created when the extension is installed. That random ID is not linked to your Google account or anything else about you.
- The off switch. The extension's popup has a switch that turns analytics off. With it off, no events are sent, and the random ID is no longer attached to requests to our API or passed to the website.
- Files you download are saved by your browser. The extension does not need or use a Samplecut account.
The Telegram bot
The Samplecut bot on Telegram (@samplecut_app_bot) takes a YouTube, TikTok, Instagram, SoundCloud or X link and sends back the audio. It needs no Samplecut account. It works like this:
- What we keep about you. We never store your Telegram user ID, name or username. The bot turns your user ID into a one-way hash made with a secret key, and only that hash is recorded, so your requests can be grouped together without your ID. Your ID is also used, in memory only, to apply the bot's limits on how many links you can send. The bot also records the language your Telegram app is set to, as Telegram reports it.
- Analytics. For each link you send, the bot records what happened to it: whether a file or a download link was sent, or why not; the format you picked; the video's platform, ID and length; and how long each step took. It also records how you found the bot, such as from the button on samplecut.app. These events go to PostHog, hosted in the EU, under the hashed ID, and are kept for one year. The hashed ID is not linked to a Samplecut account or to the website's random ID.
- Why we do this. The legal basis is our legitimate interest in running the bot, keeping it from being abused, and finding out whether people come back to it.
- Your links. To get the audio, our servers fetch the video from its platform, exactly as they do for a link pasted on the website. Like any video extracted on Samplecut, it enters our record of extracted videos and can have a public page on the site. That record is about the video, not about you.
- Files and download links. Telegram stores the files the bot sends, so we keep a reference to each one and can send the same track again without fetching it. Those references are about the video, not about who asked for it. When a track is too big for Telegram, the bot sends a download link that works for 7 days. Anyone you forward it to can use it. It is kept out of search engines, and its secret part is removed from page addresses before website analytics records them.
- Telegram. Your messages to the bot pass through Telegram, under Telegram's privacy policy.
- Deleting your data. Write to privacy@samplecut.app with your Telegram user ID (the number, not your @username). We work out its hash the same way the bot does and delete everything recorded under it.
Samplecut in ChatGPT, Claude and other chat apps
You can use Samplecut from inside a chat app, such as ChatGPT or Claude, by asking the chat to cut audio from a link. It needs no Samplecut account. It works like this:
- What the chat app sends us. When the chat uses Samplecut, the chat app sends our API the link, the part of the track you asked for and the format. It also sends the name and version of the chat app, the name of the AI model it uses, and a one-line description of what you are trying to do, written by the AI for us. We remove any links from that description before it is recorded.
- What we keep about you. ChatGPT also sends an anonymized user ID and an anonymized conversation ID that OpenAI makes for Samplecut. We never store them. Your user ID is turned into a one-way hash made with a secret key, and the conversation ID into a one-way hash, so your requests can be grouped by person and by conversation without either ID. ChatGPT also reports your approximate location and language: we keep your country and language setting, and never read or store your city or coordinates. Other chat apps send no user ID; for them our API hands the chat a random conversation ID, so the requests in one conversation can be grouped together.
- Analytics. For each request we record what happened to it: whether it worked, or why not; how long it took; the video's platform, ID and length; the part of the track and the format you asked for; and, for a link Samplecut can't cut, only the website it points to, not the link itself. When the chat app shows Samplecut's audio card, we also record whether the clip was played, re-cut or downloaded, and whether a download link was opened after it expired. These events go to PostHog, hosted in the EU, under the hashed IDs, and are kept for one year. The hashed IDs are not linked to a Samplecut account or to the website's random ID.
- Why we do this. The legal basis is our legitimate interest in running the service in chat apps, keeping it from being abused, and finding out where it fails people.
- Your links. To get the audio, our servers fetch the video from its platform, exactly as they do for a link pasted on the website. Like any video extracted on Samplecut, it enters our record of extracted videos and can have a public page on the site. That record is about the video, not about you. The download link we send back works for less than an hour, and anyone you share it with can use it.
- The chat app. Your conversation itself stays with the chat app, under its own privacy policy, such as OpenAI's or Anthropic's. We only receive the requests its AI sends to Samplecut.
- Deleting your data. You can't see the IDs a chat app sends us, so write to privacy@samplecut.app with the day you used Samplecut and the videos you cut. We find the requests that match and delete everything recorded under the same hashed IDs.
Who else handles data
- PostHog — analytics, session recordings and logs.
- Google — sign-in.
- Telegram — carries messages to and from the Samplecut bot, and stores the files it sends.
- The chat app you use, such as OpenAI's ChatGPT or Anthropic's Claude, carries your requests to Samplecut under its own privacy policy.
- Vercel — hosts the website.
- Hetzner — hosts the API and its database, in the EU.
- Cloudflare — domain and network services.
To fetch the audio, our servers make requests to the platform it comes from (YouTube, TikTok, Instagram, SoundCloud or X), sometimes through a proxy provider. Those requests come from our servers, not your browser, and do not include information about you.
Your choices
- Use Samplecut without signing in; nothing you do is tied to a name or email.
- Block analytics in the browser, or turn it off in the extension's popup.
- Ask us to delete what the Telegram bot recorded about you (see The Telegram bot).
- Ask us to delete what we recorded when you used Samplecut in a chat app (see Samplecut in chat apps).
- Ask us to delete your account and the cuts you published, or for a copy of what we hold about you, by writing to privacy@samplecut.app.
Changes and contact
If this policy changes, the date at the top changes with it. Questions go to privacy@samplecut.app.