Privacy Policy

What Reach collects, why, who receives it, how long it stays and how to remove it.

Last updated [Date]

Who we are

Reach (reachclipping.com) is run by [Legal company name], [Registered address]. In this policy, "Reach", "we" and "us" mean that company. Reach helps clippers find campaigns, cut clips from live streams, edit them, and send them to their own social accounts.

This policy covers reachclipping.com and the Reach app. By signing in to Reach you agree to this policy. Questions or complaints about your privacy go to [Contact email].

What Reach collects

From Twitch, when you sign in with it

You sign in with Twitch, so Reach never has a password for you and never sees your Twitch password.Reach asks Twitch for three permissions, user:read:email, user:read:follows and user:write:chat, and receives:

  • Your Twitch user ID, login name, display name and a link to your profile picture.
  • Your email address, if Twitch has a verified one for you.
  • The list of channels you follow.
  • Access and refresh tokens, which let Reach read the above for you, and send a message you write in Multiview's chat when you press Send. Reach doesn't keep what you send, and never sends anything to chat by itself.

From you, as you use Reach

  • Your profile: display name, username and picture. Some of it is public, as described under "Your public profile".
  • Your settings: clip defaults and notification choices, with the time zone your browser reports when you save moment alerts, so quiet hours keep your local time.
  • Channels you add on Streams, your Multiview tiles, and the clips, views and watchlists you save or hide on Trending clips.
  • Campaigns you save. Reach stores only each campaign's ID.
  • Clips you cut: the channel, the time they were cut from, their title, their video file and what was said in them.
  • Video and images you upload to the editor, your edits, and the words in your clips.
  • Links you paste to download, and what each site says about the video.
  • What you ask Reach's AI agent, what it answers, and what it cost.
  • Automations you set up, and the videos they found.
  • Computers you connect with the Reach desktop app: each computer's name and when it last sent a file.
  • API keys you make, and what apps using them read, change and ask you to approve.
  • Notifications about your posts, clips and accounts, such as a post that failed, a clip that's ready, or a post you finish by hand coming up in 15 minutes, with the reason a platform gave.
  • Posts you schedule: the caption, hashtags, title, visibility, time and made-for-kids answer, what happened to each attempt, and any post link you paste in.
  • Instagram and Facebook names you type in, for accounts you post to by hand. That's only your note of where you post, and Reach doesn't check it. Accounts you connect by signing in are covered under "Instagram and Facebook".

What Reach doesn't collect

Reach doesn't collect bank or payment details. Reach's own database doesn't record your IP address, though the companies that host Reach may keep standard request logs that include it.

YouTube

Reach uses YouTube API Services. If you connect a YouTube channel, you also agree to the YouTube Terms of Service. Google's handling of your data is described in the Google Privacy Policy.

Connecting a channel gives Reach two Google permissions:

  • youtube.upload, to upload the clips you schedule.
  • youtube.readonly, to see your channel's name and picture, and whether the videos Reach uploaded for you are live.

What Reach reads and stores: your channel's ID, name, handle and picture link, your tokens, and each uploaded video's ID, link and status. Reach doesn't read your view counts, subscribers, comments or other videos.

What Reach sends to YouTube: the clip's video file, and the title, caption, hashtags, visibility, publish time and made-for-kids answer you chose.

How it's used: only to upload what you schedule and show you what happened. Reach doesn't sell your YouTube data or use it for advertising. The only others who see it are apps you give an API key to, which can read your channel's handle, as described under "Apps you connect with an API key". Reach's use of information received from Google APIs will adhere to the Google API Services User Data Policy, including its Limited Use requirements.

How to remove Reach's access: disconnect the channel on the Integrations page, which revokes Reach's access at Google and deletes Reach's copy of your tokens. As well as Reach's own way of deleting data, you can revoke Reach's access at any time on Google's security settings page. Reach tries each connected channel once a day and marks it expired when Google says the access has ended.

Within 7 days of a channel being disconnected, or of Google telling Reach its access has ended, Reach deletes the channel's ID, name, handle and picture link, and the video IDs and links it stored for that channel. What you wrote yourself, such as your captions and titles and any post link you added, stays until you delete those clips or your account, under the label "Removed account".

Deleting data in Reach doesn't change anything stored by YouTube. To remove a video from YouTube, use YouTube.

TikTok

Connecting a TikTok account gives Reach two permissions:

  • user.info.basic, to read your TikTok ID, display name and picture.
  • video.upload, to send clips to your TikTok inbox.

What Reach stores: those profile details, your tokens, and TikTok's ID and status for each clip sent.

What Reach sends to TikTok: the clip's video file. It lands in your TikTok inbox as a draft. It isn't public until you finish it and post it in TikTok, and Reach can't see whether you do. Your caption stays in Reach for you to copy.

Reach uses TikTok information only to run this connection. It isn't sold or used for advertising, and the only others who see it are apps you give an API key to, which can read your account's name.

Disconnecting the account on the Integrations page revokes Reach's access at TikTok and deletes Reach's copy of your tokens. You can also remove Reach in your TikTok account settings. As with YouTube, Reach deletes the account's ID, name and picture link within 7 days, and what you wrote stays until you delete those clips or your account. TikTok handles your data under its own Privacy Policy.

Your public profile

Everyone with a Reach account has a public page at reachclipping.com/u/ and their username, which anyone with the link can open without signing in. It shows your display name, username, picture and the month you joined. The page never shows your connected YouTube, TikTok, Instagram or Facebook accounts, your clips or your posts. It asks search engines not to list it.

Twitch data and the 24-hour limit

Twitch's developer rules limit how long Reach can keep what Twitch's API returns, so most of it is only cached:

  • The list of channels you follow is kept for at most 24 hours. Reach refreshes it while you're using Reach.
  • Live channels you follow are read from Twitch when you look, and aren't saved to the database.
  • Twitch clips, channels and games on Trending clips are shared by everyone. Twitch has to confirm each one within 24 hours or it's deleted, and view history is cut at 24 hours.
  • A clip you saved or hid is held only by its ID, and is removed if Twitch says the clip has gone.

This limit doesn't cover your own account details from Twitch, which stay until you delete your account, or channels you add on Streams. Your link to a channel on Streams stays until you remove it. The channel's public details, such as its name and picture link, stay in a table shared by everyone, and aren't tied to you.

Live streams and clip files

Multiview records public Twitch and Kick streams on Reach's private recorder while someone is watching them, and stops soon after nobody is. It keeps a rolling 5 minutes so a clip can be cut, and deletes older footage.

The clips you cut are stored in Cloudflare R2 and stay there until you delete them or your account. They can include the streamer's picture and voice. If you're in a clip made on Reach and want it looked at, contact [Contact email].

Uploads and the editor

Videos you upload, from your browser or the desktop app, go straight from your computer to Cloudflare R2 through a link that works for 6 hours. Images you upload for your brand are stored on Reach's server. An upload you start but never finish is deleted after 24 hours.

Reach's editor works on its own server, called the worker. It keeps copies of the videos you edit there, and makes previews, thumbnails and exports from them. Finished exports are stored in Cloudflare R2.

To frame a facecam, the editor looks for faces in a few frames of the clip. That runs on the worker with a model that comes with Reach, and no picture is sent anywhere. Reach keeps only where to crop, not the face.

Transcripts, titles and AssemblyAI

Reach uses AssemblyAI to turn what's said in a clip into text. It does this at four moments:

  • When a clip you cut in Multiview finishes, so it can have a title from what was said.
  • When you open a clip or an upload in the editor, for captions, unless it already has its text.
  • When you ask the editor to find moments in a longer video.
  • When an automation brings in a new video, for the whole video.

The worker takes only the audio from that stretch and sends it to AssemblyAI, with names and game terms saved for that channel. Those word lists are shared by every clipper who edits the channel. The picture isn't sent. The audio can include the voices of anyone in the clip.

AssemblyAI returns the words, their timing, who spoke and the language. Reach keeps the words with the clip and with your editor project until you delete them or your account. When a transcript finishes, Reach asks AssemblyAI to delete it. If a transcript fails partway, Reach doesn't send that request. AssemblyAI handles the audio under its own Privacy Policy.

Titles are picked from the transcript by Reach's own code, and no AI model writes them. A title from a transcript never replaces one you gave the clip yourself.

Downloading from links

When you paste a link in the editor, other than a Twitch or Kick stream Reach records itself, Reach's worker reads it and downloads the part you pick, from YouTube or any other site, with yt-dlp and FFmpeg. The site sees Reach's server asking, not your computer. Before reading a link, Reach checks that it points to a public site.

  • Reach keeps the link only while that download runs, and clears it when the download finishes, fails or is cancelled. A link you look up but don't download stays with that lookup until it's a day old, and is deleted the next time you paste a link.
  • Reach keeps what the site said about the video: its title, channel, length, qualities and picture address. It also keeps the quality and part you picked, and a key that recognises the same video again, so it can offer you the file it already has.
  • The file is saved as one of your uploads, in Cloudflare R2, under the same rules as your other uploads. Deleting it in the editor deletes the download's record too.
  • The record of a failed download, without its link, stays until you delete your account. You can clear it while it's listed, which is for a day.
  • The worker gets yt-dlp updates from GitHub.

Instagram and Facebook

Once Meta has approved Reach, you can connect an Instagram professional account or a Facebook Page and Reach posts for you. Until then, or whenever you'd rather, Reach gets the clip to your phone and you post it yourself.

  • Connecting Instagram gives Reach instagram_business_basic and instagram_business_content_publish, to read your account's ID, username, name and picture, and to publish the Reels you schedule.
  • Connecting Facebook gives Reach pages_show_list, pages_read_engagement and pages_manage_posts, to list the Pages you manage, read each Page's ID, name and picture, and publish the Reels you schedule.

Reach stores those details, your tokens, encrypted, and each post's ID, link and status. It sends Meta the video file and your caption with its hashtags. Reach doesn't sell this or use it for advertising, and the only others who see it are apps you give an API key to, which can read your account's name.

Disconnecting the account in Integrations deletes Reach's copy of your tokens, and its details go within 7 days, as with YouTube. If you remove Reach in Instagram's or Facebook's own settings, or ask Meta to have your data deleted, Meta tells Reach. Reach then disconnects the account and deletes what came from Meta straight away, and gives Meta a confirmation code to show you.

The AI agent

The editor has an AI agent that can write post text, fix a clip so it meets a campaign's rules, and answer what you ask it about a clip. It runs on Anthropic's Claude: Claude Haiku first, and Claude Sonnet as a second try when a rule fix or an answer from Haiku doesn't work. It only runs when you press its button or send it a message, and it's off when Reach has no key for Anthropic.

What it sends to Anthropic: the clip's title, length and transcript with timings, its captions, layout and post text, the campaign's rules, the channel the clip came from, and what you typed. It never sends pictures, audio or video, and never sends your name, email, accounts or any ID that says who you are.

What Reach keeps: each request, its answer and the edits it made, with the editor project, until you delete that project or your account. A record of what each request cost stays until you delete your account. Each clipper has a daily and monthly spending limit. Anthropic handles what it receives under its own Privacy Policy.

Sound cleanup

To balance voice, game sound and music, or to clean up a voice, the editor runs sound models on Reach's own worker. No audio leaves Reach's computers for this. The separate tracks it makes stay with the editor project on the worker, and go when you delete the project, the clip or your account.

Automations

An automation watches for new videos for you and turns them into clips, held for you to review in the editor.

  • For a YouTube channel, Reach checks the channel's public feed about every 15 minutes. It uses no YouTube sign-in or access of yours for this.
  • For a Twitch channel, Reach checks when a stream ends and downloads its past broadcast.
  • New videos are downloaded on Reach's worker, as a pasted link is, and stored as your uploads. Their audio goes to AssemblyAI to find the moments, as described above. Automations never use the AI agent.
  • Reach keeps each automation, what it found, and the clips it held, until you delete them or your account. Deleting an automation keeps the clips it already held.

The Reach desktop app

The desktop app is another way to download links, on your own computer. It downloads videos from links you paste, and sends the ones you choose to Reach's editor. Its downloads run on your computer with yt-dlp and FFmpeg, straight from the site the video is on. Reach's servers don't download those, and the app doesn't use Reach's access to YouTube. The video stays in your Videos/Reach folder, or a folder you pick, until you delete it.

Connecting a computer: the app sends Reach your computer's name and opens Reach in your browser, where you approve it while signed in. The code lasts 10 minutes. Reach then stores the computer's name, when it last sent a file, and a hash of its token. The token itself stays on your computer, encrypted by your operating system.

What it sends to Reach: only the video files you choose to send, and their titles. It never sends the link a video came from.

What stays on your computer: your downloads, and a list of them with each link, title and file location.

Other services: each time it starts, the app checks GitHub for a newer yt-dlp and downloads it from GitHub. It loads thumbnails from the sites your videos come from. It has no analytics or crash reporting.

Disconnecting: disconnecting in the app, or removing the computer under Connected computers in Settings, deletes Reach's record of it straight away. If the app can't reach Reach at the time, it says so and stays connected. A computer that sends nothing for 30 days stops working and has to be connected again. Once it has gone 60 days without sending anything, Reach deletes its record the next time anyone connects a computer or opens Connected computers.

Apps you connect with an API key

In Settings you can make API keys for other apps and AI tools, including ones that connect over MCP. A key acts as you, with read or edit access.

  • Reach shows a key once, and stores only a hash of it and its first few characters. Each key ends after 7, 30 or 90 days, and you can revoke it sooner.
  • An app with a key can read your profile, clips, editor projects, exports and links to download your files, your posts and their captions, your connected accounts' names, the streams and campaigns you'd see in Reach, and your stats. With edit access it can also make changes, such as renaming a clip or starting an export.
  • Posting, moving, cancelling or deleting a post, deleting clips and submitting to a campaign need your approval in Reach. Reach keeps each request, with a summary of what it would do, until you delete your account. A request you don't answer within 3 days ends.
  • Reach logs what each key does, as IDs and outcomes, never titles or captions. To let you undo an edit, it keeps the old value. Log entries older than 30 days, and old values older than 7 days, are cleared each time you open the activity log.
  • Reach keeps its answer to each change an app makes, so a repeated request can't act twice. Answers older than 24 hours are cleared the next time that key makes a change.
  • For an app connected over MCP, Reach keeps the name and version the app gives itself. That goes once the connection has been unused for 30 days, the next time any app connects over MCP.

When a YouTube or TikTok account's details are removed, as described above, its name is also taken out of these requests and answers. What an app does with what it reads is up to that app and its own privacy policy, and Reach can't take it back.

How Reach uses your information

  • To sign you in, and keep you signed in.
  • To show live channels you follow, their popular clips and trending clips.
  • To send the Twitch chat messages you write, when you press Send.
  • To record, cut, store, title and edit the clips you ask for.
  • To run the AI agent when you ask it, and automations you set up.
  • To download the links you paste, and receive the files you send from a computer you connected.
  • To let apps you give a key to read and act for you, within what you allow.
  • To send your clips to the accounts you chose, at the time you chose.
  • To check that your connections still work, and show you what happened to each post.
  • To make sure a platform account belongs to only one Reach login.

Reach doesn't sell your information, and doesn't use it to show you ads.

Who receives your information

  • Twitch, YouTube and TikTok receive what's described above, because you connected them.
  • AssemblyAI receives clip audio for transcripts.
  • Anthropic receives what the AI agent works on, as described above, when you use it.
  • Meta receives the clips and captions you post to Instagram or Facebook.
  • Cloudflare stores clip files, uploads, exports and live video in R2.
  • GitHub is where Reach's worker and the desktop app get yt-dlp updates. It receives nothing about you from Reach.
  • Sites you paste links from receive the request for that video from Reach's worker.
  • Apps you give an API key to receive what that key can read, as described above.
  • Vercel runs the website and app. Hetzner runs the recorder. [Database and editor host] runs the database and the editor.
  • Anyone with a link you create. Send to phone makes a link that needs no sign-in. A link for a scheduled post lasts 7 days, and a link for an exported clip lasts 24 hours. Share them only with people you trust.
  • Authorities, if the law requires it.

Campaigns run on Content Rewards, a Whop product. Reach's server reads its public campaign pages, and the rules pages on reachclipping.com they link to, and sends them nothing about you. When you join a campaign, submit a clip or get paid, you're using Content Rewards, under Whop's privacy policy.

Cookies and your browser

Reach places these cookies on your browser:

  • reach_session keeps you signed in. It holds a random token, and Reach stores only a hash of it. It lasts 30 days, is renewed as you use Reach, and ends 90 days after you signed in.
  • reach_oauth, and reach_social_oauth protect a sign-in while it's happening, and last 10 minutes.
  • reach_rail remembers whether the sidebar is collapsed, for a year.

Reach also keeps small things in your browser's own storage, such as notices you've dismissed, your editor setup and your place in a page. The Reach app has no advertising or analytics cookies.

Some content on Reach comes straight from other companies, which can see your IP address and may set their own cookies:

  • Twitch serves the clip player on Trending clips, and may show its own content there, including ads.
  • Twitch serves channel pictures and stream previews. Content Rewards serves campaign images.
  • Twitch, 7TV, BetterTTV and FrankerFaceZ serve the badge and emote pictures in Multiview's chat. Reach's server reads their public lists of a channel's emotes, which sends them the channel, not anything about you.

Reach's public campaign pages use Plausible Analytics to count page visits, clicks on outbound links and clicks on buttons such as Join Campaign. Plausible says it uses no cookies and doesn't store IP addresses. The Reach app doesn't use Plausible.

How long Reach keeps things

  • Your account, profile and settings: until you delete your account.
  • Your Twitch tokens: until you delete your account, or Twitch ends Reach's access.
  • YouTube and TikTok tokens: until you disconnect the account or delete your Reach account.
  • A disconnected YouTube or TikTok account's ID, name, handle and picture link: up to 7 days.
  • Clips and their files, transcripts, posts and post history: until you delete them or your account.
  • Editor projects, runs, uploads, brand images and exports: until you delete them or your account.
  • Uploads you started but never finished: 24 hours.
  • Saved campaigns: until you unsave them or delete your account.
  • A pasted link: until its download ends, or a day for a lookup you didn't download, as described above.
  • API keys: a key that ended or was revoked more than 60 days ago is deleted the next time you open your keys. App requests: until you delete your account.
  • Connected computers: until you remove them or delete your account. One that sends nothing for 30 days stops working, and its record goes after 60. A connection code lasts 10 minutes.
  • Notifications: 30 days, and no more than 50 at a time. One about something still waiting on you, such as a failed post, stays until that's dealt with.
  • Clips you hid on Trending clips: 31 days.
  • Twitch data with a time limit: as described above.
  • AI agent requests and answers: with their editor project, until you delete it or your account. What they cost: until you delete your account.
  • Automations, what they found and the clips they held: until you delete them or your account.
  • The recorder's rolling footage: 5 minutes.

Your choices and deleting your data

  • Change your details in Settings.Your email comes from Twitch, so change it there and sign in again.
  • Disconnect YouTube or TikTok on the Integrations page.
  • Remove Reach from Twitch in your Twitch connections. Reach then signs you out everywhere, clears your Twitch tokens and deletes your follow list when it next checks with Twitch, which it does about hourly while you're using Reach.
  • Remove a connected computer under Connected computers in Settings.
  • Delete your account in Settings.

Deleting your account takes effect straight away. Reach revokes its access at Twitch, YouTube and TikTok, and deletes your account and everything tied to it from its database. That includes your profile and public page, your ways to sign in, tokens, sessions, connected accounts, connected computers, clips, transcripts, posts, notifications, editor projects, downloads, API keys, app requests and saved items. It then deletes your files from storage: clip recordings and exports, uploads, including unfinished ones, brand images, and the editor's working copies and previews.

Deleting a single clip, editor project, run or brand image deletes its files the same way. An upload goes with the last project or run that used it. Once a day, Reach also looks for files that nothing in Reach uses any more and deletes them.

Some things are different:

  • Facecam framing and word lists you set for a channel are shared with other clippers. They stay, with your name taken off.
  • Deleting data in Reach doesn't delete anything from Twitch, YouTube, TikTok, Instagram, Facebook or Content Rewards. Posts on those platforms have to be removed there.

For anything else about your data, contact [Contact email].

Security

Twitch, YouTube and TikTok tokens are encrypted with AES-256-GCM before they're stored. Each one is tied to its own account, so a token copied to another row won't decrypt, and tokens never reach your browser. A YouTube upload link in progress is encrypted the same way. The cookie that keeps you signed in can't be read by scripts on the page, and is only sent over HTTPS.

No system is perfectly secure, so we can't promise your information will never be accessed without permission.

Children

Reach isn't aimed at children. You must be at least [Minimum age] to use it.

Changes to this policy

When this policy changes, the date at the top changes too. If Reach starts to use your data in a way this policy doesn't cover, we'll ask you to accept the new version first.

Contact

[Legal company name]
[Registered address]
[Contact email]

See also the Terms of Service.