This policy covers horrac.com, your Horrac account and Ostrel, the desktop app we make (formerly called Gecko). Horrac is the trading name of Rikraj Haldar ("Horrac," "we," "us"), who is responsible for your data. It explains what we collect, what Ostrel accesses on the platforms you connect, where that data goes (including the parts that run on servers we operate, not just on your own device), and how you can remove it.
The short version
When you connect a YouTube, Twitch, Instagram, or TikTok account, Ostrel talks to that platform's own servers using an authorization you grant through their official sign-in screen. Some of what Ostrel does happens locally, on your own device. Some of it happens on a small backend we operate (we call it "the relay"): keeping your credentials so background checks can run, generating Channel Reports and anomaly alerts automatically, and staging content for scheduled publishing. It needs to keep working even while the app itself is closed, which is why it lives on a server rather than only on your device. We never sell your data and never use it for advertising. The app does send anonymous, feature-level usage events (never your content, credentials, or anything identifying you personally) to a product-analytics provider so we can see which parts of the app are actually useful and when something breaks; you can turn this off at any time in Settings. See "Anonymous usage analytics" below for exactly what that involves. Separately, and only if you turn it on, Ostrel can send a masked replay of how you move around the app, with every word, number, image and video hidden on your computer first; see "Screen replays" below.
What the app accesses, and why
| Platform | Scopes requested | What each is used for |
|---|---|---|
| YouTube | youtube.readonly, yt-analytics.readonly | Your channel name, subscriber count, and your own video/analytics data (views, watch time, traffic sources, audience demographics). The same data YouTube Studio shows you. |
| YouTube | youtube.upload | Only if you use the scheduling/publishing feature. Lets Ostrel upload a video to your channel on your explicit instruction (you choose the file, caption, and schedule). Ostrel never uploads anything without you initiating that specific action. |
| Twitch | moderator:read:followers | Your channel name and follower count. Your past broadcasts, clips and live status (with viewer counts) are read from Twitch's public API, which needs no extra permission, for stream stats, clip insights and your Channel Report. |
instagram_business_basic, instagram_business_manage_insights | Your account name, follower count, and your own posts' reach/engagement numbers. | |
instagram_business_content_publish | Only if you use the scheduling/publishing feature. Lets Ostrel publish a post to your account on your explicit instruction. | |
| TikTok | user.info.basic, user.info.stats, video.list | Your TikTok account ID, display name, profile picture link, and follower, following, total-likes and video counts; and your own videos' IDs, captions, posting dates, lengths, cover images and view/like/comment/share counts. |
| TikTok | video.publish | Only if you use the scheduling/publishing feature. Lets Ostrel publish a video to your account on your explicit instruction. Until our TikTok integration passes TikTok's own audit, anything published this way is visible only to you (SELF_ONLY); that's TikTok's own restriction, not one Ostrel imposes. Before each post, Ostrel also reads your TikTok nickname, username and profile picture link, the privacy options and comment, duet and stitch settings your account allows, and the longest video it accepts, to build the posting screen and check the schedule. It doesn't save these, apart from the privacy level the post was sent with. |
None of these scopes grant access to private messages, other users' data, payment information, or the ability to read or modify your account settings. The publishing scopes let Ostrel post on your instruction. They never let Ostrel delete, edit, or read content you didn't create through the app, and Ostrel never publishes automatically without you reviewing and confirming what's being scheduled first.
Ostrel's use and transfer to any other app of information received from Google APIs adheres to the Google API Services User Data Policy, including the Limited Use requirements.
YouTube. Ostrel uses YouTube API Services to read your channel's data and, if you schedule posts, to upload to it. Google's own use of your data is covered by the Google Privacy Policy. You can revoke Ostrel's access to your YouTube account at any time on Google's security settings page, myaccount.google.com/permissions, or disconnect YouTube in Ostrel's Settings. Using Ostrel with YouTube also means agreeing to the YouTube Terms of Service.
Where your data is stored
Ostrel has two places data can live, and which one depends on what you're using:
- Locally, on your device. Your settings, your local cache of connected-account data, and anything you haven't opted into the automated/background features stays only in your own device's application-data folder. Your account credentials are encrypted at rest locally too, not stored as plain text, using a separate encryption key from the server-side one described below.
- On our servers (the relay), for features that need to keep working while the app is closed. Specifically: your account credentials (encrypted at rest, see below), your Channel Reports and content-anomaly history, competitor-tracking data, cadence/hashtag insights, the measured and transcript facts about your videos and the links between your videos and your projections and Reviews (see "Measuring your videos and checking results" below), and anything you schedule for future publishing (the media file itself, temporarily, plus its caption and timing). Generating a weekly Channel Report or firing off a scheduled post can't depend on your local device being open at that exact moment, so this data has to live somewhere that stays running.
Your data follows your account, not your computer. When you sign in on a new computer, or after reinstalling, the history kept on our servers (Channel Reports, content alerts, projections, reviews and other synced app data) is copied to that computer from the one you used most recently, so it isn't lost. Platform connections aren't copied; you reconnect them on the new computer.
Signing out keeps your history on your account. Signing out removes Ostrel's data from that computer, but your platforms stay connected to your Horrac account: background checks keep running, scheduled posts still go out, and everything is there when you sign back in. If you'd rather end all of that, tick "Also disconnect my platforms" when you sign out, which cancels Ostrel's access and deletes that computer's server-side data, or use "Disconnect all platforms" on your account page at horrac.com/account. If a signed-out computer's data has been kept for 180 days and your account no longer has an active plan or trial, we delete it.
Credential encryption. Access tokens and any platform-app secrets we store server-side are encrypted at rest using AES-GCM, with the encryption key held separately from the database itself.
Nothing about your accounts or content is shared with third parties beyond what's already described above (the platform APIs themselves, and the AI/search providers described in "AI features" below). The one addition is anonymous, feature-level usage analytics, and the opt-in masked screen replays, each covered in its own section below, neither of which carries your account data, content, or anything identifying you.
Data security
All traffic between the app and our servers is encrypted in transit (TLS). Server-side and local credentials are encrypted at rest as described above. Our database provider (Cloudflare) keeps hourly backups for 24 hours and offers point-in-time recovery for up to 30 days, for disaster-recovery purposes only. So that the service can be restored even if our Cloudflare account were lost, we also keep a weekly copy of that database with Google Cloud Storage for up to 28 days. The copy is made by an automated job on GitHub, encrypted there before it is stored, and can only be decrypted with a key we keep offline. We don't use either kind of backup to work around a deletion request, but it means data can theoretically be recoverable by us for up to 30 days after deletion, for that limited purpose. Our servers also keep a short technical log of the requests they handle (for example which feature was used, a reference to the account or installation, and any error) for up to 7 days, to keep the service running and to investigate faults. An error in it can occasionally include part of a request that failed. If we become aware of a data breach affecting your personal data, we'll notify the ICO where required by law and, where there's a real risk to you, notify you directly without undue delay.
Reporting a security issue. If you believe you've found a security vulnerability in Ostrel or our backend, email support@horrac.com with details. Give us a reasonable opportunity to investigate and fix the issue before disclosing it publicly, and we won't pursue legal action against good-faith, non-destructive security research reported this way.
Your Horrac account
To use Ostrel you sign in to a Horrac account, on horrac.com or in the app. What that involves:
- Your email address is what identifies the account. We store it encrypted (AES-GCM), and look it up only by a keyed hash (HMAC-SHA256), so the address itself is never used as a database key and a copy of the database alone would not reveal it.
- Sign-in codes are sent to that address through Resend, our email provider. We keep only a hash of each code, it expires after 10 minutes, and it can be used once.
- Signing in with Google gives us your verified email address and nothing else from your Google account.
- Creating your account. When you create the account you give us a name to call you by (up to 40 characters, stored encrypted like your email, and changeable any time on your account page), and we record the date you accepted these terms and this policy. You also make two choices, both of which you can change later on your account page: whether to get occasional product news by email (off unless you tick it), and whether Ostrel shares anonymous usage data (see "Anonymous usage analytics"; the choice is saved to your account so it applies on every device).
- Unfinished sign-ups. If you sign in but never finish creating the account, we delete it, with its sessions, after 7 days, unless a payment is linked to it.
- Sessions. Signing in creates a session so you stay signed in. In the app it is stored encrypted on your device.
- Cookies on horrac.com. Only three, all strictly necessary and sent only to horrac.com over HTTPS.
__Host-horrac_sessionholds your session; only our server can read it, and it expires after 30 days without use.__Host-horrac_incontains only the value "1", so the page can show your account menu instead of "Sign in"; it has the same lifetime and carries nothing about you.__Host-horrac_googleexists only while you sign in with Google, holds one-time security values for that sign-in and expires after 10 minutes. horrac.com uses no advertising, analytics or tracking cookies. - Visit counts on horrac.com. The website uses Cloudflare Web Analytics to count visits: which page was opened, the site you came from, your country, your browser and device type, and how fast the page loaded. It sets no cookies, does not fingerprint your browser and does not identify you, and the counts are never linked to your Horrac account. Cloudflare processes this for us as part of hosting the site.
- Credits. To apply your plan's limits, we record for each AI request which feature used it, which provider and model it went to, how many tokens it took and the resulting number of credits. We don't keep the content of your requests for this purpose.
- Abuse limits. To stop sign-in codes, billing pages and free trials being misused, we briefly count requests (and free trials started) per email address (by its keyed hash), per account and per network (your IP address; for IPv6, only the network part). These counts are deleted within two days, and used sign-in codes within a day of expiring.
- One free trial per computer. When a trial starts, the Ostrel app sends a one-way hash of your Windows machine ID (never the ID itself), and we store only a keyed hash of that, with the date. It can't be turned back into the ID or linked to your computer outside Horrac. We keep it after an account is deleted, so deleting and re-creating an account can't restart the free trial.
- After you delete your account we keep only its random account ID and the date, with no email address or other detail, for 90 days. That lets us recognise and cancel a payment made for the account after it was deleted, for example in a checkout left open in another tab.
Payments
Paid plans are bought through Stripe. Purchases are sold through Link, Stripe's merchant-of-record service, which also handles sales tax and VAT. Stripe collects your payment details and billing address directly; we never receive your full card number. We store only your Stripe customer and subscription IDs, which plan you are on, its status and its renewal date. Stripe and Link process your payment information under their own privacy policies as independent controllers, and keep transaction records for as long as the law requires them to, even after you delete your Horrac account.
Automated background checks (Channel Report, Content Anomaly, Cadence Insight)
For platforms where this is enabled, Ostrel periodically checks your connected account's public performance data on our servers, even while the app is closed, to keep your Channel Report current and flag unusual video/post performance. This uses the same read-only data described in the table above. It never posts, comments, or changes anything on your account. You can see exactly when a report last ran, and you can delete a platform's stored report data at any time (see Data Deletion below).
Anonymous usage analytics
Ostrel sends a small number of feature-level events (e.g. "a report was generated," "a chat reply was sent," "an account connection succeeded or failed") to PostHog, a product-analytics provider, so we can see which parts of the app are actually used and where people get stuck, especially during onboarding. This runs entirely from the app's own background process, never from the screens you see and interact with, and it does not include any recording or replay of your screen. Screen replays are a separate choice, off unless you turn them on, described in the next section.
We also send an event when the app hits an unexpected internal error (e.g. an uncaught exception), tagged only with the general kind of error. The details of that error go in a separate error report, described in "Automatic error reports" below.
What's sent, and what isn't:
- Events are tagged only with which feature was used and basic structural context (e.g. which platform), tied to a random installation identifier that already exists for this copy of the app. They are never tied to your name, email, or any platform account.
- Events never include your content, chat messages, report text, credentials, error messages, stack traces, or any of the data described elsewhere in this policy.
- Now and then Ostrel may ask a one-tap question about something you just saw, for example whether a result told you something new. Your answer is sent as one of these events, carrying only the option you picked. There is never a box to type into.
- This is on by default. You can turn it off when you create your account ("Help improve Ostrel"), and at any time in the app's Settings under "Share anonymous usage data" or on your account page. The choice is saved to your Horrac account so it follows you to every device. Turning it off stops all future events; it does not retroactively delete anything already sent.
- PostHog stores these events in its EU region.
Screen replays (opt-in)
If you turn on "Share how you use Ostrel" (in Settings, or when Ostrel first asks), Ostrel sends PostHog a masked replay of how you move around the app, so we can see where a screen is confusing or slow and make it better and more responsive. It is off unless you turn it on.
What a replay contains, and what it doesn't:
- The layout of the screens you open, where you click and scroll, where you pause, and how long screens take to appear.
- Every word and number on screen is replaced before it leaves your computer, so we see grey shapes, not your titles, captions, chat, reports or analytics. Anything you type is hidden the same way. Pictures, thumbnails and videos are replaced with empty boxes, and their addresses are removed.
- No console logs or network requests are recorded.
- Replays are tied to the same random installation identifier as the usage events above, never to your name, email or any platform account.
While replays are on, the top bar shows "Replay on", and one click stops it. You can also turn it off in Settings at any time. Turning it off stops future replays; it does not delete replays already sent. Replays are deleted automatically after 30 days. To have them deleted sooner, email us at the address below.
PostHog stores replays in its EU region.
Logs you choose to send
Ostrel keeps a log on your computer: its own technical record of what it did, such as a report that failed or a platform that refused a request. It stays on your computer unless you click Send log in Settings. Then Ostrel sends us the most recent part of it so we can work out what went wrong, and shows you a short reference to mention when you contact us.
Before it leaves your computer, Ostrel removes your home folder name, email addresses and anything that looks like a password, key or token, and our server checks again before storing it. Error messages in the log can still mention a video title or other text from the request that failed. The log is linked to your installation of Ostrel, stored on our backend (Cloudflare), never shared with anyone else, and deleted automatically after 30 days.
Automatic error reports
When Ostrel hits an unexpected error, such as a crash or a feature that stops working, it sends us a short error report: what went wrong, where in Ostrel's own code it happened, and which versions of Ostrel and Windows you use. It goes through the same cleaning as a log you send: Ostrel removes your home folder name, email addresses and anything that looks like a password, key or token, and our server checks again before storing it. An error message can still mention a video title or other text from the request that failed. Error reports are linked to your installation of Ostrel, stored on our backend (Cloudflare), never shared with anyone else, and deleted automatically after 30 days. They follow the "Share anonymous usage data" setting, so turning that off stops them too.
AI features (Channel Report, Suggestions Chat, Pre-Publish Critique, the Projection Model, Media Kit, Recurring Patterns)
These features send the relevant content, meaning your account's public content/performance data described above, and (for video-based features) a small, representative sample of your own videos, to Google's Gemini API or Anthropic's Claude API through our relay, using a credential we own, never one you provide. Specifically:
- YouTube and Instagram: a small sample of your videos (never your entire library) may be sent for AI analysis. YouTube videos are referenced by link; Instagram videos are downloaded from Instagram's own servers and uploaded to Google's Gemini Files API through our relay, where Google retains them for at most 48 hours before automatic deletion. Other content sent to Google's Gemini API (prompts and the replies to them) is logged by Google for up to 55 days, solely to detect and prevent misuse. Content sent to Anthropic's Claude API is retained for up to 30 days under our commercial agreement with them, and is not used to train Anthropic's models.
- TikTok: TikTok offers no lawful way for a third-party app to download video, so Ostrel never takes a video from TikTok. By default Ostrel sends each TikTok's text metadata (captions and hashtags, dates, lengths, view/like/comment/share counts and the engagement rate worked out from them, and each video's ID) and its cover image, the still TikTok's own API provides, so the AI can judge the cover. Ostrel fetches the cover fresh for each analysis and does not keep a copy. If you want Ostrel to watch a TikTok, it uses your own copy of that video instead: the same video you posted publicly as a YouTube Short or Instagram Reel (Ostrel matches these by caption, length and posting date, and you can undo any match), the original file of a TikTok you posted through Ostrel's scheduler, or a file you add from your computer. That copy is sent for AI analysis the same way as the YouTube and Instagram videos above. A file from your computer is compressed on your device and uploaded straight to Google's Gemini Files API, where Google deletes it within 48 hours. The file itself stays on your computer; Ostrel only remembers where it is, and you can remove it from Ostrel at any time. Which of your YouTube or Instagram posts each TikTok is (and which it isn't) is kept with your synced history, so these matches follow your account to a new computer; the locations of files never leave your computer.
- Twitch: Twitch offers no lawful way for a third-party app to download video, so only text metadata (titles, dates, view counts) is sent. No Twitch video or audio ever leaves your device.
- Transcription sends the sound of a video to Groq, a hosted transcription provider, for speech-to-text conversion: a draft you review (Pre-Publish Critique uses your own words to find strong clips and moments), and the first 10 minutes of your own recent posts that Ostrel measures (see "Measuring your videos and checking results" below). Groq doesn't use this audio to train its own models, on any tier, and doesn't otherwise retain it beyond a short troubleshooting/abuse-prevention window.
Every call to Google's Gemini API or Anthropic's Claude API uses a paid, billing-enabled key on our side, never a free one. Both providers' paid tiers contractually exclude submitted content from being used to train their models, which a free tier would not guarantee. AI-generated output (reports, critiques, projections, suggestions) is not reviewed by a human before being shown to you, and should be treated as a starting point, not a guarantee. See the Terms of Service for more on this.
Trend and topic search. For Instagram and TikTok, some features (Suggestions Chat, the Projection Model) ground their answers in a search of current trends or events on the open web. This normally goes through Google's own search-grounding tool as part of a Gemini call. On the rare occasion our daily grounding budget for that is exhausted, we fall back to Parallel.ai's Search API instead, sending only a short search topic or query, never your personal account or content data.
Measuring your videos and checking results
So that Diagnosis and Review rest on measured facts rather than guesses (Ostrel uses them to check what its AI writes, even where it doesn't show them), and so Ostrel can learn which of its advice and projections hold up for you, Ostrel does the following. All of it is about your own videos and is used only for you: it is never shown to anyone else or combined with other creators' data.
- Measuring, on your computer. Ostrel counts the cuts, pauses, first sound and loudness of your recent posts and of drafts you review, with software that runs on your computer, not an AI service. It measures from a copy it may use: the file Ostrel posted for you, your own copy of a TikTok, a file you added, a file in an export folder you chose for Ostrel to watch, or one you dropped on Ostrel (it reads the names, dates and lengths of the videos there to find the right one, on your computer only; when you click Find my videos, it reads the same details in your Videos, Desktop, Downloads and Documents folders, your synced cloud folders and the top folders of your other drives to suggest a folder, and skips files kept online only so nothing is downloaded; when a new video appears in a watched folder, Ostrel may offer to Review it, and that notification never includes the file name), the draft you reviewed in Ostrel that a post was made from, or an Instagram video downloaded from Instagram and deleted from your computer straight after measuring. YouTube videos are never downloaded: a YouTube video is measured only through the file Ostrel posted, the draft you reviewed in Ostrel, or the same video on another platform. The measured facts (not the videos) are stored on your computer and on our servers, so reports made there use them. For YouTube they include how many viewers were still watching at each point, from YouTube's own analytics.
- Sound, picture and word fingerprints of reviewed drafts. When you review a draft video, Ostrel keeps a fingerprint of its sound (a list of numbers describing its pattern of tones, which cannot be turned back into sound), of its picture (for each second, a 128-bit summary of how brightness changes across a 9 by 9 grey thumbnail, which cannot be turned back into the picture) and of its words (each run of five words in its transcript turned into a number, which cannot be turned back into the words). They stay only on your computer, for up to 90 days, so Ostrel can recognise a posted video made from that draft by comparing the two, including when the music was changed or the shots were reordered or cropped. A posted video's sound and picture fingerprints are worked out on your computer for that comparison and not kept. Its word fingerprint is kept on your computer for up to 6 months, so a draft transcribed later can still be compared and so phrases you say in many videos (an intro or a sign-off) are not taken as a match. When you review a thumbnail, Ostrel keeps the same kind of 128-bit picture summary of the thumbnail, only on your computer, and compares it with the thumbnails of the YouTube videos you post next, so it can recognise the video that used it. A posted video's thumbnail is fetched from YouTube for that comparison, summarised on your computer and not kept. Deleting the Review deletes its draft's fingerprints.
- Transcript facts. After transcription (see "AI features" above), Ostrel keeps only facts about the words: how fast you speak, the first line and the first 60 words, and when the topic and the payoff are first said, which a short text request to Anthropic's Claude API finds. The full transcript is not kept. Audio waiting to be transcribed stays on your computer and is deleted once transcribed.
- Links and results. Ostrel notes which posted video came from which projection or Review: certainly when you posted the same file through Ostrel, otherwise from the sound, the picture, the words, the edit or what the video shows. It then records how each linked video did two weeks and a month after it went up (views, likes, comments and, on YouTube, how many viewers were still watching at 30 seconds), and, for a Review, which of its tips you used. An automatic link is only Ostrel's judgement: you can remove it with "Not this video". Links, results and grades are stored on your computer and on our servers, without where your files are on your computer.
Disconnecting a platform deletes its measured facts, transcript facts and links, on your computer and on our servers, and the word fingerprints of its videos on your computer (for YouTube, after the same 7 days as its other data, or straight away if its access was removed). Deleting a Review deletes its fingerprints, transcript facts, links and grades. Deleting your account deletes all of it.
Scheduled publishing
If you use Ostrel's scheduling feature, the media file you select (and a custom thumbnail, if you set one), its caption or title, and its scheduled time are uploaded to our servers so the post can go out at the right time even if the app is closed. Once a scheduled post is published (or you cancel it), that staged media and thumbnail are deleted from our servers. If a post fails for good (it could not be published and retrying won't help), they are deleted within a day of the failure, or eight days after upload if that is later. The post's record stays on our servers until you delete your account: its caption or title, scheduled time and platform, the settings you chose for it, the ID the platform gives the post, and any error it reported. Ostrel uses it to show you what happened to each post and, for TikTok, to match a posted video to the file you posted. We also email you once about each post that fails for good, through Resend, with the post's title (or its file name, or the start of its caption), its platform, its scheduled time and the error the platform reported, so you hear about it even when the app is closed. You can turn these emails off on your account page. Nothing is published without you first reviewing it and confirming the schedule.
Our legal basis for processing your data
Under UK GDPR Article 6, we process your data on these bases:
- Performance of a contract (Art. 6(1)(b)), for the core functionality you've asked us to provide: your Horrac account and sign-in, connecting your accounts, showing your analytics, generating reports, publishing content you've scheduled and telling you when a scheduled post fails, applying your plan's credit limits, and billing.
- Legitimate interests (Art. 6(1)(f)), for automated background checks (Channel Report refreshes, content-anomaly detection) that run on our servers even while the app is closed. Our interest is delivering the always-current monitoring the app promises. We've weighed this against your right to control it: you can disable it or delete a platform's stored data at any time (see Data Deletion).
- Legitimate interests (Art. 6(1)(f)), for anonymous usage analytics (see "Anonymous usage analytics" above). Our interest is understanding which features are actually useful so we can improve the app. We've weighed this against your right to control it: the data is anonymous (tied only to a random install identifier, never your account or content), and you can turn it off at any time in Settings.
- Legitimate interests (Art. 6(1)(f)), for automatic error reports and our servers' short technical log (see "Automatic error reports" and "Data security" above). Our interest is finding and fixing faults before they reach more people. We've weighed this against your right to control it: reports are cleaned of names, emails and keys before they leave your computer, are deleted after 30 days (the server log after 7), and you can turn error reports off with "Share anonymous usage data".
- Legitimate interests (Art. 6(1)(f)), for keeping the service from being abused: sign-in and billing rate limits, and the one-free-trial-per-computer check. Our interest is keeping sign-in secure and the free trial fair; the data involved is a keyed hash, never your IP address history or your machine ID itself.
- Consent (Art. 6(1)(a)), for product-news emails, which we send only if you ticked that box. You can withdraw it at any time on your account page, and every such email has an unsubscribe link.
- Consent (Art. 6(1)(a)), for a log you choose to send from Settings (see "Logs you choose to send" above), which we receive only when you click Send log.
- Consent (Art. 6(1)(a)), for screen replays (see "Screen replays" above), which are off unless you turn them on. You can withdraw it at any time in Settings or from the top bar.
- Nothing else relies on consent. For the platform data above, disconnecting an account (see Data Deletion) stops us processing that platform's data going forward.
International data transfers
horrac.com and Ostrel's backend run on Cloudflare, its weekly encrypted backup is made on GitHub and stored with Google Cloud, AI processing runs on Google and Anthropic, audio transcription runs on Groq, and sign-in and service emails are sent through Resend. These providers process data outside the UK, mainly in the United States, so using them involves an international transfer of personal data. The safeguard differs by provider:
- Cloudflare, Resend and GitHub are certified under the UK Extension to the EU-U.S. Data Privacy Framework, which the UK government recognises as providing adequate protection.
- Google, Anthropic, Resend and Groq: their data-processing agreements with us include the EU's Standard Contractual Clauses with the UK's International Data Transfer Addendum, or the ICO's International Data Transfer Agreement, the UK's standard contractual safeguards for transfers to a country without its own adequacy decision.
- PostHog stores our anonymous usage events and the opt-in screen replays in its EU region, which UK law recognises as adequate.
- Stripe and Link handle payments as independent controllers and are responsible for their own transfers under their own privacy policies.
- Parallel.ai receives only a short search topic, never personal data.
The ICO has indicated both UK transfer mechanisms may be updated during 2026. We'll keep this section current as that happens. If you have questions about a specific transfer, contact us at the address below.
Requests from public authorities
We only disclose user data to police, government agencies or regulators when we are legally required to. Before responding to any request, we check that it is genuine and has a valid legal basis. We disclose only the minimum data the request legally requires, and where a request appears unlawful or broader than necessary, we will seek to challenge or narrow it. We keep a record of every request we receive and how we responded. Where the law allows, we will tell the affected user before disclosing their data.
Your rights
Under UK GDPR Articles 15 to 22, if you're in the UK or EEA you have the right to:
- Access the personal data we hold about you, and rectify it if it's inaccurate.
- Request erasure ("the right to be forgotten"). See Data Deletion below for the practical mechanism.
- Restrict or object to certain processing, including the legitimate-interests-based background checks described above.
- Receive a copy of your data in a portable, machine-readable format.
- Rights relating to automated decision-making. These don't apply here: every AI-generated output (a report, a critique, a projection) is a suggestion you review and choose whether to act on. Ostrel never makes a decision about you, or automatically takes an action with legal or similarly significant effect, without your own review and confirmation first.
You also have the right to complain to the UK Information Commissioner's Office (ico.org.uk) if you believe we've mishandled your data. We'd appreciate the chance to address it directly first, at the contact below.
These are your statutory rights under UK GDPR if you're in the UK or EEA. If you're elsewhere, we'll still honor equivalent access, correction, and deletion requests as a matter of practice. See Data Deletion below.
Data deletion
See our Data Deletion page for how to disconnect an account, delete server-side report/schedule data, or delete your whole Horrac account. Deleting your account from the app's Settings or from your account page on horrac.com cancels any subscription and erases your account and every device's server-side data in one step.
Children's privacy
Ostrel's Terms of Service require you to be at least 18, or the age of majority in your jurisdiction, to use it. Ostrel is not directed at children, and we do not knowingly collect data from anyone under that age.
Changes to this policy
If this policy changes, we'll update the "Last updated" date above. We'll tell you about material changes in the app before they apply.
Contact
Questions about this policy: support@horrac.com