Terms of Service Privacy Policy Cancellation & Refunds Imprint

Privacy Policy

Version 2026-10-13.2 · Last updated: October 13, 2026 · GDPR / DSGVO

The short version: atomAI is local-first. Your chats, notes, memory and videos live as files on your PC — not in our database. Our servers only know what is needed to run your account. We show no ads, set no tracking cookies and use no analytics scripts, and we never sell data.

1. Controller

The controller within the meaning of the GDPR is the provider named in the Imprint. Contact: [email protected].

2. What we process, why, and on which legal basis

2.1 Website visits

When you visit atomai.pro, our web server processes technical connection data (IP address, time, requested page, browser type) in server logs to deliver the site and to detect abuse (Art. 6 (1)(f) GDPR — legitimate interest in a secure service). Logs are deleted or anonymised after 14 days at the latest. Our site sets no cookies for visitors and loads all fonts and scripts from our own domain. The only thing it keeps in your browser is your choice of light or dark mode, if you make one.

How we count website use. We measure how our website is used on our own server — without any script in your browser and without cookies. For each page that is opened we keep one line with the time, the page, the website the visitor came from (only its domain) or the page of ours they came from, campaign tags contained in the link (such as utm_source), the country (as reported by Cloudflare, Section 3) and whether a phone or a computer was used (from the browser identification), and whether the request came from a person's browser or from an automated program (judged only from the technical information every browser sends with each request anyway), so that we count people, not bots. The same applies to downloads of our installers. To count a visitor only once per day, the server derives a code from the IP address and browser identification using a random key that changes every day; this code is held only in memory for that day and is never stored. We also count sign-ups, verified accounts, started checkouts, purchases (product, tier and amount, without the account), support and bug reports. No IP address, no name and no account is stored with any of these counts, so they cannot be traced back to you; they are deleted after 120 days. If you reach us through an advertisement, the campaign tags in that link (which campaign and advertisement, the search term and whether a phone or a computer was used) are passed on to our product, order and download links and noted in the counts described above; if you buy, they are also kept with the order at our payment provider (Section 3), so that we can see which advertisement led to a download or a purchase. Nothing is stored in your browser for this; with “Do Not Track” or “Global Privacy Control” switched on, the tags are not passed on. If your browser sends “Do Not Track” or “Global Privacy Control”, your visits are not counted. Our apps contain no analytics and no tracking. Legal basis: Art. 6 (1)(f) GDPR (our legitimate interest in understanding and improving our website and offers).

Purchases that come from a Google ad. Google adds a click identifier (gclid) to the links of its ads. What we do with it depends on where you are. If you visit from the European Economic Area, the United Kingdom or Switzerland — or if we cannot tell where you are — we do not use it: our order page drops it, and apart from the ordinary server log described above it is not stored and not passed on to anyone. For buyers from other countries, and only if the billing country of the payment is also outside those countries, we keep the identifier with the purchase in our own records (not at our payment provider) and make the following available for Google Ads to collect: the click identifier, the time of the order, the amount and currency, and an order number of our own. No name, no email address, no IP address and no account. Google uses this to attribute the purchase to the ad (Section 3). Refunded purchases are not reported. We delete the click identifier 63 days after the purchase, and after 21 days if no paid purchase followed. Legal basis: Art. 6 (1)(f) GDPR (our legitimate interest in knowing which advertisement leads to purchases). You can object at any time (Section 5); with “Do Not Track” or “Global Privacy Control” switched on, the identifier is not used at all.

2.2 Account

If you register, we process your email address, your password as a salted hash (never in clear text), your plan and account settings (Art. 6 (1)(b) GDPR — performance of contract). You sign in with your email address. Each account also has an internal account name — the username you chose if your account was created before October 2026, otherwise one derived from your email address. Signed-in areas use one strictly necessary session cookie; it contains no tracking and expires when the session ends. To protect accounts we limit repeated sign-in attempts by IP address and tell you by email when your account is signed in to from a new device or browser (Art. 6 (1)(f) GDPR — security of the service).

Sign in with Google or Discord (optional). If you choose this instead of a password, the provider sends us your user ID at that provider, your email address and your display name, and we store them to create or find your account and to keep the sign-in working. We request no further permissions for signing in. You can unlink the provider at any time in the app as long as another way to sign in remains. Legal basis: Art. 6 (1)(b) GDPR.

After a purchase. So that we can help when a setup gets stuck, we note on your account whether and when you downloaded an app’s installer (only when you are signed in or come from your purchase confirmation page — the download links in our emails carry no identifier), when that app first and last signed in with your account, and whether it was unlocked or why not (for example a wrong password). The apps report nothing extra for this; these are steps that pass through our server anyway. Legal bases: Art. 6 (1)(b) GDPR (delivering what you bought) and Art. 6 (1)(f) GDPR (our legitimate interest in being able to help). These notes are deleted together with your account.

2.3 Proof of consent and declarations

When you accept our Terms (at registration), start a paid plan (including the consent to immediate performance under the withdrawal rules), or declare a cancellation, a withdrawal or a refund request, we record what was declared, the document version, the timestamp, your IP address and technical context. Legal basis: Art. 6 (1)(c) and (f) GDPR — we are required to be able to demonstrate consent (Art. 7 (1) GDPR). These records are kept for as long as statutory proof and limitation periods require.

2.4 Emails

We send transactional emails only (confirmation, password reset, important account or legal notices) from [email protected] (Art. 6 (1)(b) GDPR). No marketing emails without separate consent.

Sending yourself a download link. If you open the Scribe download on a phone or tablet, you can have the link emailed to you. We use the address you type for that one email only: it is not saved, not linked to an account and not used for advertising. To keep the form from being misused, our server remembers for up to 24 hours, in its working memory only, that an email went out to that address, and limits how many emails one connection can request. Legal basis: Art. 6 (1)(b) GDPR (your request) and Art. 6 (1)(f) GDPR (protection against misuse).

Support and bug reports. If you write to us — by email, through the support form or with a bug report from the app — we process what you send (your message, your email address and name, the app version and where in the app it happened, and for the support form your IP address to prevent misuse) to answer you and to fix the problem. When our support changes something on your account at your request, we keep a record of what was changed and by whom, and may add a short note to your account. Legal basis: Art. 6 (1)(b) GDPR, and Art. 6 (1)(f) GDPR where no contract is involved.

2.5 App data — local first

The desktop app stores your workspace (chats, memory, notes, projects, calendar, upload drafts) in Documents\Atom on your device. Videos you edit in atomAI are processed locally and are not uploaded to us. We have no access to these files. Apart from the shared niche pool described next, the app contacts our servers only for sign-in, plan status, updates and the platform features you actively use. Optional components (the local AI runtime, speech recognition, the video tools) are downloaded by the app directly from their publishers (currently GitHub, Hugging Face and Ollama, all USA), which see your IP address as with any download. If you connect a third-party service with your own key (for example stock footage or voice-over), the app sends your requests to that service directly from your PC, under your own account with it.

Shared niche pool. The Niche finder searches YouTube from your PC, not from our server. So that you do not start with an empty screen, the app sends the public channel and video figures it just found (channel name and ID, title, views, subscribers, upload frequency) to a shared pool that every atomAI user reads from, and it does this in the background when the app starts. This contains no data about you: not your account, not your searches as a profile, not your own channel unless it happens to be a public search result. Entries expire automatically. Legal basis: Art. 6 (1)(f) GDPR (our and our users’ legitimate interest in a usable niche database built from public platform data).

2.6 Connected platforms (YouTube, TikTok)

If you connect a channel, we retrieve the data needed for the feature you use (e.g. public channel statistics, video metrics, comments) and cache it briefly (currently 10 minutes) to respect API quotas (Art. 6 (1)(b) GDPR). If a feature requires OAuth authorisation (e.g. uploads or private analytics), access tokens are stored encrypted and used only for the requested features. You can disconnect at any time in the app; for Google also via the Google security settings.

Some features (e.g. “Look up a channel”, Trends) also show publicly available statistics of channels you don’t own. We retrieve this public data on your request and cache it briefly; legal basis: Art. 6 (1)(f) GDPR (our and your legitimate interest in showing public platform statistics).

atomAI uses YouTube API Services. By using YouTube features you also agree to the YouTube Terms of Service. Google’s processing is described in the Google Privacy Policy.

2.7 Google user data — how we use it and with whom we share it

This section applies when you connect your Google/YouTube account to atomAI. It describes exactly which Google user data we access, why, and — as required by Google — with whom we share, transfer or disclose it.

What we access (only after you grant it on Google’s consent screen):

Where it is processed. Your OAuth access and refresh tokens are stored encrypted on our server in the EU (IONOS, Germany) and are used solely to make the API calls for the feature you triggered. The data we retrieve (your statistics, videos, revenue figures) is shown in the app and cached on your PC. Video files you upload go directly from your PC to YouTube — they are never uploaded to, stored on or transited through our servers.

Sharing, transfer and disclosure of Google user data. We do not sell Google user data, we do not use it for advertising or profiling, and we do not use it to develop, improve or train AI/ML models. The AI in atomAI runs locally on your own PC; Google user data is never sent to any external AI provider. We do not transfer Google user data to any other third party, with these narrow exceptions:

No other person, company or service receives your Google user data. We do not share it with other atomAI users, and we do not publish it.

Limited Use. atomAI’s use and transfer of information received from Google APIs to any other app will adhere to the Google API Services User Data Policy, including the Limited Use requirements.

Retention and revocation. You can disconnect Google at any time in Settings → Connections — this deletes the stored tokens on our server immediately. You can also revoke access at any time in your Google account permissions. Tokens are deleted when you disconnect, when you delete your atomAI account, or when they become invalid. Legal basis: Art. 6 (1)(b) GDPR (performing the features you request).

2.8 TikTok user data — how we use it and with whom we share it

This section applies when you connect your TikTok account to atomAI. It describes exactly which TikTok user data we access, why, and with whom we share, transfer or disclose it.

What we access (only after you grant it on TikTok’s authorisation screen):

Where it is processed. Your OAuth access and refresh tokens are stored encrypted on our server in the EU (IONOS, Germany) and are used solely to make the API calls for the feature you triggered. The data we retrieve (your profile figures and your videos’ numbers) is shown in the app and cached on your PC. Video files you post go directly from your PC to TikTok — they are never uploaded to, stored on or transited through our servers.

Sharing, transfer and disclosure of TikTok user data. We do not sell TikTok user data, we do not use it for advertising or profiling, and we do not use it to develop, improve or train AI/ML models. The AI in atomAI runs locally on your own PC; TikTok user data is never sent to any external AI provider. We do not transfer TikTok user data to any other third party, with these narrow exceptions:

No other person, company or service receives your TikTok user data. We do not share it with other atomAI users, and we do not publish it.

Retention and revocation. You can disconnect TikTok at any time in Settings → Connections — this deletes the stored tokens on our server immediately. You can also revoke access in your TikTok account settings. Tokens are deleted when you disconnect, when you delete your atomAI account, or when they become invalid. Legal basis: Art. 6 (1)(b) GDPR (performing the features you request).

2.9 Payments

Paid plans and purchases are processed by Stripe Payments Europe, Ltd. (Ireland/EU). Your card or payment details are entered on Stripe’s secure checkout and never reach our servers; we receive only what we need to run your subscription (plan, status, billing period, a customer reference). If you buy without an account, Stripe also passes us the email address you entered at checkout, and we create your account with it. We are your contracting party for the purchase; Stripe is our payment processor. Stripe’s processing is described in the Stripe Privacy Policy. Legal basis: Art. 6 (1)(b) GDPR. Stripe may transfer data to its parent company in the USA; this is covered by the EU standard contractual clauses and Stripe’s certification under the EU-US Data Privacy Framework.

2.10 Keeping the service fair (anti-abuse)

To enforce plan limits and protect the service against manipulated app copies, the desktop app sends along with its server requests: a random installation ID (created on your PC; it identifies the installation, not you personally), the app version, a technical integrity checksum of the app’s own program files (never of your content or workspace), and your plan’s usage counters (e.g. AI credits used). Per installation we also keep a shortened code derived from the network address it connects from — not the address itself. We use this to keep usage consistent across your devices, to notice when one account is used by several people, and to detect modified app copies; a modified copy may be limited to the Free plan. No workspace content (chats, notes, memory, videos) is involved. Legal bases: Art. 6 (1)(b) GDPR (enforcing the agreed plan) and Art. 6 (1)(f) GDPR (preventing abuse). If your app is ever restricted and you believe this is a mistake, contact [email protected] — a person will review it.

2.11 Discord community & referrals (optional)

If you link your Discord account, we store your Discord ID and username to connect your atomAI account with our Discord server (e.g. plan or creator roles). Community leaderboards are strictly opt-in per board: only if you enable them do we show your channel name and the relevant figures (e.g. views; revenue only as a tier, with your consent) on the server. Our referral programme records who joined through your invite to credit rewards. You can unlink Discord and disable each board at any time in the app; unlinking removes the roles. Legal basis: Art. 6 (1)(b) GDPR, and Art. 6 (1)(a) for the leaderboards.

2.12 atomAI Scribe (dictation app)

Your speech stays on your device. Scribe recognises speech with a model that runs on your own PC. Your audio, your dictated text and your history never reach us. There is no telemetry and no analytics in the app; besides your sign-in, the app contacts us only for the activation check described below and to look for updates (it fetches a small version file from atomai.pro, which, like any download, sees your IP address; nothing about you or your dictation is sent). Setting it up downloads the speech model once from the open-source project at GitHub (Microsoft Corporation, USA), which sees your IP address as with any download.

Activation. To unlock the app, Scribe sends us the unlock file it holds, a random device ID created on your PC and the product name, and checks back every few days. The activation service stores a one-way hash of your account reference, the device IDs using it with their first and last day of use, the device name your device reports (so your device list is readable instead of a row of random IDs — this can be a name you set, such as “Anna-Laptop”), and your tier — no email address and no IP address. All of this is deleted when you delete your account (for Standard, see also “Activation check for Standard” below). Since 21 September 2026 the unlock file carries that same hash, so an unlock stays tied to the account that bought it; unlocks issued before then have no such reference and keep working unchanged. This is what limits an unlock to the devices you bought, and you can see and release those devices in your account. When a device is signed out — because you released it, you changed your password, you deleted your account, the Standard activation check found it shared, or there was no free slot — we keep a short marker (the random device ID, the reason and the time — no name, no IP address) for up to 45 days, so the device can tell you why it was signed out; once it expires the rest is removed too. Legal basis: Art. 6 (1)(b) GDPR.

Activation check for Standard. Minimum and Standard are for one person (Terms, Section 3). To detect whether one Standard purchase is used by several people at the same time, the activation check that Scribe performs while you dictate also processes technical connection data. This data is only held temporarily for the check and then discarded; it is not stored permanently. If the check indicates shared use, a device can be signed out, and you can sign in again at any time. This leaves the same short sign-out marker described under “Activation” (random device ID, reason and time — no IP address), kept for up to 45 days so that a person can review it. Legal bases: Art. 6 (1)(b) GDPR (a Standard purchase is for one person) and Art. 6 (1)(f) GDPR (our legitimate interest in preventing misuse). You can object at any time (Art. 21 GDPR) by writing to [email protected] — a person will look at it.

Purchase check. Since 21 September 2026 we sell Scribe ourselves, and your purchase is held with your atomAI account; payment runs through Stripe (Section 2.9) and your card details never reach us. Purchases made through Gumroad before that date are still verified against Gumroad, Inc. (USA), which handled that payment under its own privacy policy; transfers to the USA are covered by the EU standard contractual clauses.

3. Recipients and processors

We never sell personal data and we do not use it for profiling. We use it for advertising in one case only: the purchase report to Google Ads described in Section 2.1. We do not use automated decision-making within the meaning of Art. 22 GDPR, with one narrow exception: a technically modified app copy can automatically be limited to the Free plan (Section 2.10) — you can always have this reviewed by a person via [email protected].

3a. Marketing emails (only if you ask for them)

What. If — and only if — you tick the optional box when creating your account (or turn it on later in your account), we store your email address, the fact and time of your consent, your IP address and browser at that moment (proof of consent, Art. 7 (1) GDPR), which of our emails you were sent, and whether you later unsubscribed.

How consent works. Ticking the box is not enough on its own: we send you one confirmation email, and only when you click the link in it do we consider you subscribed (double opt-in). Until then we send you no marketing at all.

Legal basis. Your consent (Art. 6 (1)(a) GDPR, § 7 (2) no. 2 UWG). We deliberately do not rely on the “existing customer” exemption of § 7 (3) UWG — we only mail people who actively asked us to.

Withdrawal. Every marketing email contains a one-click unsubscribe link that works without logging in and takes effect immediately, and your mail program may also show its own unsubscribe button. You can also write to [email protected]. Withdrawing costs nothing and has no effect on your account.

What this never covers. Emails about your account itself — password resets, security notices, billing, cancellation confirmations — are not advertising. They are necessary to perform the contract (Art. 6 (1)(b) GDPR) and cannot be switched off while you have an account.

Service announcements. We may occasionally email all users — regardless of the marketing setting — about genuinely important operational matters: planned downtime, a security incident, or an essential change to the Service, your plan or these terms. These are not advertising; they are part of providing the Service to you (Art. 6 (1)(b)/(f) GDPR) and therefore have no unsubscribe. We do not use this channel for offers or promotions.

Sending. Marketing emails are sent through our email provider IONOS (see Section 3). We do not track whether you opened an email and we do not use tracking pixels.

4. Retention

Account data is kept while your account exists and deleted promptly after account deletion. Residual copies can remain in our server backups, which are access-restricted and are deleted after 30 days at the latest. Usage counters, installation IDs and community/leaderboard entries are deleted together with the account, including the Scribe activation record (device IDs, device name and tier). Connection data for Scribe’s Standard activation check is only held temporarily for the check; sign-out markers (device ID, reason and time — no name, no IP address) are kept for up to 45 days. Support requests and bug reports are kept only as long as needed to handle them and are deleted on request. Click identifiers from advertisements (Section 2.1) are deleted after 63 days at the latest. Consent records and billing records are kept for the statutory proof and retention periods. Everything stored locally on your PC is under your control at all times.

5. Your rights

You have the right to access (Art. 15), rectification (Art. 16), erasure (Art. 17), restriction (Art. 18) and data portability (Art. 20). Where processing is based on your consent, you can withdraw it at any time with effect for the future (Art. 7 (3)). Contact [email protected]. You may also lodge a complaint with a supervisory authority (Art. 77 GDPR) — in Germany, the data-protection authority of your federal state.

Right to object (Art. 21 GDPR). Where we process your data on the basis of legitimate interests (Art. 6 (1)(f) GDPR), you can object at any time for reasons arising from your particular situation. We then stop that processing unless we have compelling legitimate grounds for it or need the data for legal claims. An email to [email protected] is enough.

Do you have to give us data? An email address and a password (or a Google or Discord sign-in) are needed to have an account, and payment details are needed to buy — without them we cannot provide the account or the purchase. Everything else is optional.

6. Security

All connections are encrypted in transit. Passwords are stored only as salted hashes, platform tokens are stored encrypted, and our server holds none of your working data — your workspace stays on your device.

7. Changes

We update this policy when the Service or the law changes. The current version always applies; material changes are announced by email or in the app.