Privacy Policy
Last updated November 17, 2026
This policy explains what THE FANGATE collects, why, and your choices.
What we collect
Account details (name, email, workspace). Content you create or upload. Connected-platform tokens (stored encrypted). Usage and analytics data. Billing details are handled by our payment processor — we don’t store full card numbers. If you add photos of yourself to build a character sheet, you provide likeness media, which is sensitive personal data.
How we use it
To operate the Service: generate and schedule content, publish to the platforms you connect, compute analytics, meter usage, and provide support. We don’t sell your personal data.
Third parties we share with
AI model providers (to generate content), the social platforms you connect (to publish), our payment processor (to bill), and infrastructure providers (to host). Each receives only what’s needed for its function. Some may process data outside your country; where required, we rely on appropriate safeguards.
By name, these are the companies that receive personal data, and what each one gets:
OpenRouter runs our text models and routes each request to the model’s own provider (today DeepSeek, Google, MiniMax or NVIDIA). It receives the text you type, your brand voice, text we fetch from a product page you link, the content of a support email so a reply can be drafted, and photos when you ask a model to describe them. The section on AI providers says which of this may be kept and trained on.
Google Cloud (Vertex AI and Cloud Text-to-Speech) renders storyboards, character sheets, stills, video and music, and narrates scripts. It receives the images and footage you upload for editing or video, your photos and approved character sheet with each shot’s instructions, your scripts and the briefs behind them. The search terms behind title and SEO suggestions go to the YouTube Data API. Data from a connected YouTube channel is covered in its own section below.
Cloudflare generates images with Workers AI from the prompts we build, and holds the off-site copies of our encrypted backups in R2 (see Security and retention).
Perplexity receives your brand or product name and the questions we ask about it when you run the AI-search visibility check.
Stripe bills you. It receives your email address, your name, your workspace’s name and id, and the address you give at checkout. Card details go to Stripe only, never to us.
DigitalOcean hosts the server that runs the Service, with its database and media, and takes a weekly image of the whole server disk.
Sentry receives error reports from our server and from the app in your browser: the error, its trace, and the request’s method and path. The request body, cookies, query strings, Authorization headers and the values held in variables are stripped before a report is sent.
Resend delivers the email we send you (verification, password reset, invitations, support replies): the address, the subject and the message. Forward Email relays mail sent to support@thefangate.com to our mailbox at Google (Gmail) and to the Service, so a copy of every support message you send sits with those two as well as with us.
The platforms you connect (YouTube, TikTok, Instagram, Facebook, X) receive what the next section describes.
Connected platforms: what we read and write
For every account you connect on the Channels page we store its platform id, display name, handle and profile-picture link, the access tokens the platform issued (encrypted at rest and never returned by any API), the permissions it granted, when it was connected, and the figures we later read about the posts we published there. Each platform grants only what is listed for it below.
YouTube. Described in full under Google user data, below.
TikTok. TikTok gives us your account’s id, display name and profile picture, and permission to publish videos to it. We read no figures from TikTok: the permission to list your videos has not been granted to the app, so TikTok posts show no numbers.
Instagram. You sign in with Instagram itself; no Facebook Page is involved. Instagram gives us your professional account’s id, username, name and profile picture; permission to publish images and video to it; and read access to its follower and media counts and to the reach, likes, comments, saves and shares of the posts we published.
Facebook Page. Facebook gives us the list of Pages you manage (including Pages owned through a business portfolio) so you can pick one; permission to post to that Page’s feed, photos and videos; and read access to the Page’s follower count and to the likes, views and insights of the posts we published.
How Instagram and Facebook results steer your own content. Besides showing you how your posts did, the results we read from your Instagram account and your Facebook Page, and any figures you enter by hand, are used to steer what THE FANGATE makes for your workspace: the platform, posting hour, title pattern, format and audience that have done best for you are written into the instructions we send to AI model providers, described under AI providers and model training below. A pattern learned this way acts only after you approve it. These results are not used to steer any other workspace’s content. Results from YouTube, TikTok and X are never used this way.
X. Described in full under X user data, below.
Disconnecting a channel on the Channels page wipes its tokens and cancels anything not yet sent through it; the record of what was published stays. If you remove THE FANGATE from your Facebook or Instagram account, Meta tells us and we disconnect the channels you connected the same way. If you ask Meta to delete your data, we also clear the profile Meta gave us, delete the analytics readings taken through the channel and the Meta post ids on your posts, and record the request with a confirmation code whose status you can check. Meta’s request is matched to channels by the account id Meta sent us when you connected; if none match, the record says so.
Google user data (YouTube)
When you connect a YouTube channel, Google gives us, with your consent, your channel’s identity (name, ID, thumbnail), the ability to upload videos to that channel and set their title, description and privacy on your behalf, and read access to the performance of what we published. We use this only to publish the posts you approve and to show you how they performed. We share, transfer or disclose Google user data to no one except our hosting provider (DigitalOcean), which stores it encrypted on our behalf; Cloudflare, whose R2 storage holds our encrypted nightly backups (a backup contains the database, so it contains the YouTube tokens and figures the database holds); and Google itself when we make the API calls you asked for. We do not pass it to AI model providers, advertisers, data brokers or any other third party; we do not sell it; and we do not use it to train models or to serve ads. Disconnecting the channel on the Channels page deletes the stored token, and you can revoke our access at any time at myaccount.google.com/permissions. THE FANGATE’s use and transfer to any other app of information received from Google APIs will adhere to the Google API Services User Data Policy, including the Limited Use requirements. Google’s own handling of your data is described in the Google Privacy Policy.
X user data
When you connect an X account, X gives us, with your consent, the account’s name, username, ID and profile image; permission to publish posts and upload pictures or video to that account on your behalf; and read access to the public metrics of the posts we published there and of the account (followers and number of posts). We publish to X only posts you have written or approved, to one X account per post, at the time you choose; autopilot never posts to X. We use the metrics to show you how your posts performed and which of your own X posts did best. The tokens X issues are stored encrypted. No data from X, whether a figure, a post, an account detail or a pattern learned from your X results, is sent to an AI provider or used to steer what THE FANGATE writes, and we do not use X data to train or fine-tune a model. We read a post’s X metrics for 30 days after it is published and then delete them. If X tells us a post we published there no longer exists, we delete its link and the figures we read from it. Disconnecting the account on the Channels page deletes its stored tokens. You can also revoke THE FANGATE’s access from your X account’s settings at any time.
AI providers and model training
The models behind the Service are run by third-party providers under our accounts; you never contract with them directly. To do its job, the Service sends them what you give it: the text you type (briefs, product descriptions, brand voice, audience notes), text we fetch from a product page you link, images or footage you upload for editing or video, scripts for narration, and the photos and character sheets used to put you in generated stills or video. If you email support, the content of that email is also sent to a model so a reply can be drafted for our team to review before anything is sent back to you.
Some of that text goes to free model endpoints that may keep it and train on it. Part of our text generation runs on providers’ free-tier endpoints. Those providers may retain prompts and completions and may use them to train their models, and we do not control that retention. We have chosen to allow this routing; it is part of how the plans are priced. If we change that choice, this section will change with it. Image, video and likeness providers are governed by their own retention policies, and likeness media is handled as described below.
Because of this, do not put passwords, API keys, payment details, other people’s personal data, or material you must keep confidential into a brief, an upload, or a support email. Anything you would not paste into a public form does not belong in a prompt.
Likeness / biometric data
Putting yourself in a video starts with photos of you that you upload on the Likeness & products page, the Reels pages or the setup wizard. Before the first photo is added you are asked to confirm whose likeness it is and that its use is consented to, and to say where you live; we keep a record of that answer. Your photos are then sent to the image-generation provider to build a character sheet — one image with several views of the same person under one light — which you approve before it is used. Your photos and the approved sheet are sent again, with each shot’s instructions, to that provider when it renders the stills and video; you do not choose or contract with that provider — we do. There is no talking-head product. You must have consent to use any likeness. Treat this as the most sensitive data the Service handles. Removing a photo or a look on any of those pages deletes its file from our storage, and removing a photo also deletes the character sheets built from it; stills and video already made stay until you delete them. You can also ask us to delete any stored likeness media; the backups described below are the last copies to go.
Security and retention
Platform tokens are encrypted at rest; passwords are hashed. We keep data while your account is active and as needed for legal, billing, and operational purposes, then delete or anonymize it. The periods:
Your uploads, character sheets and generated content: until you delete them or the workspace. Narration audio is cached for up to 30 days in a cache shared across workspaces, so a deleted workspace’s narration lines can remain for up to 30 days; composited video is kept for 14 days; a data export for seven days. Records of each generation, of model routing, of alerts and of the email we sent you: 12 months. The record of channel and member changes in a workspace: 24 months. A tracked link’s referring host and device class: 90 days. X figures: 30 days, as above. Support mail: kept with no set period in our records, and in our mailbox at Google and at Forward Email per their retention; it is not removed when a workspace is deleted or an account is erased. Server logs: capped at 50 MB per container; the app’s go with its container at each deploy. Error reports: per Sentry’s retention.
Encrypted nightly backups are kept for 14 days on our server; after three nights a backup’s archive of media files is removed from the server, once its off-site copy is confirmed, and only its database part stays there for the rest of the 14 days. Each is encrypted on the server with a public key before it leaves it; the private key is held off the server. A copy of each set goes to Cloudflare R2 the same night, where it is kept for the period set in that account; none of our scripts deletes an off-site copy. DigitalOcean also takes a weekly image of the whole server disk, kept per DigitalOcean’s retention; our backup encryption does not cover it. A deletion reaches those copies only when they are removed.
Your rights
Depending on where you live, you may have rights to access, correct, export, or delete your data, and to object to certain processing. A workspace owner can take a copy of the workspace’s data from Settings, under Your data: one file, ready within minutes, kept for seven days. A workspace owner can also delete the whole workspace from Settings, under Delete workspace: it removes the workspace, its content, its connected channels and its media for every member, once any active subscription has been cancelled from Billing. To erase your user account itself, or for correction or an objection, write to support@thefangate.com.
Visitors who are not signed in
When you open our public pages we count the view per day, page and source (the utm_source in the address, else the site that sent you, else “direct”); the count carries no address and nothing about you. The pages load two typefaces from Google Fonts, so your browser requests them from Google. Our server writes one access-log line per request: the network address it came from, the page requested with any query string removed, your browser’s identification and the response; Authorization and Cookie headers are dropped. If your browser blocks something under our content-security policy, it posts a report; we log the page, the directive and the blocked origin, never a path or query, and keep the directive and origin in one record that each new report overwrites. If the app fails in your browser, an error report goes to Sentry as described above. When you click a tracked link in a creator’s post (an address under /go/), we record the time, the host of the site you came from and whether the device was a phone, a desktop or a bot, and pass a click id to the destination; the host and device class are deleted after 90 days. Nothing here uses a cookie.
Cookies
We use only what’s necessary to keep you signed in and run the app. We don’t use third-party advertising cookies. Signing in stores your session tokens in your browser’s local storage, not in a cookie. Before you sign in, your browser’s session storage for that tab holds the utm tags of the address you arrived on and the host of the site that sent you, sent once with registration and otherwise gone with the tab, and a sentence you type into the landing page’s try-it box so the sign-up can carry it over.
Contact
Privacy questions or requests: support@thefangate.com.
