Privacy Policy
Last updated:
FreeAIText is a static site with a small server component. This policy describes what happens to your data today, section by section, and flags separately what would happen under a feature that is built but not yet active.
The short version: the text you analyse or clean never leaves your browser. That is true whether or not you have an account, and it stays true when you buy credits — an account holds an email address and a balance, never a document. Nothing on this site currently sends your text to Anthropic, because Anthropic's watermark detection API has not been released.
1. Local browser processing
This covers the analyser and the editor as they work today.
- The text you paste or upload is read into your browser's memory and processed there by JavaScript.
- It is not transmitted to us, not written to any server, and not stored beyond the tab you are using. Closing or refreshing the tab discards it.
- The watermark estimate, the segment scoring and the synonym substitution all run locally.
- With the “online synonym dictionary” switch on, the tool sends individual single words — never sentences, paragraphs or files — to the public Datamuse API. Those requests are governed by Datamuse's own terms. Switch it off and the tool makes no outbound content requests at all.
- A single entry in local storage remembers your light/dark theme preference. No tracking or advertising cookies are set.
2. Anthropic API processing
Future-policy preview. This describes a feature that is built but not active. It is not a statement of current processing.
This is not happening today. Anthropic has not released its public watermark detection API, FreeAIText is not connected to it, and no text you submit is sent to Anthropic. This section describes what would apply if and when that integration goes live, so you can read it before deciding to use the feature rather than after.
If an Anthropic API mode is enabled in future:
- The interface will say so before submission, and will ask for your explicit consent on each run. Consent will never be assumed, remembered as a preference, or bundled into another action.
- The text you submit for analysis would be transmitted over TLS to Anthropic's detection endpoint, through this site's own server so that the API key is never exposed to your browser.
- We would store the analysis result and its metadata only as long as needed to display it to you in that session. We would not build a corpus of submitted documents.
- Anthropic's own handling of text sent to its API would be governed by Anthropic's terms and privacy policy, which we would link here once published. We cannot describe them in advance, and will not guess.
- This section will be rewritten as a statement of current processing, with a new “last updated” date, before any such feature is switched on.
3. Accounts and sign-in
An account is optional. Analysis works without one. You only need to sign in to run a cleaning pass, because that is the paid action.
Signing in is an email address and a one-time code — there is no password and no third-party account. What we store for an account is:
- your email address, lower-cased and trimmed;
- an opaque account identifier;
- your credit balance and the ledger of how it changed;
- when the account was created and when it was last used.
No name, no profile, no password. Sign-in codes are stored only as a keyed
hash and are single-use, so a copy of our database contains nothing that can be
used to sign in as you. Session cookies are HttpOnly,
Secure and SameSite=Lax, are not readable by
JavaScript, and are not used for tracking.
Documents are never attached to an account. Text you analyse or clean stays in your browser and is not written to your account record, your credit ledger, or anything else.
To delete your account and everything attached to it, contact us at the address in the footer.
4. Payment data
Credits are bought through PayPal. Card and PayPal account details are entered on PayPal's own pages. They are never seen by, sent to, or stored on FreeAIText — we never receive a card number.
For each purchase we store:
- the PayPal order and capture identifiers;
- which credit pack was bought, the amount, and the currency;
- the status of the purchase and when it completed;
- the resulting entry in your credit ledger.
PayPal's own handling of your payment is governed by PayPal's privacy statement, not by this policy.
We keep purchase records for as long as tax and accounting rules require, which is longer than we keep anything else. That is the one category of data we cannot delete on request while the obligation lasts.
5. Waitlist data
This applies now, to the launch-notification form on the Anthropic API status page. Signing up is entirely optional.
If you sign up, we store exactly this and nothing else:
- your email address, lower-cased and trimmed;
- the timestamp at which you gave consent;
- the signup source (which page the form was on);
- an optional campaign code, if you arrived from a link carrying one;
- the notification status — whether the launch email has been sent, and whether you have unsubscribed;
- a random unsubscribe token, so that no unsubscribe link ever contains your email address.
Text you paste into the analysis tool is never stored with a waitlist signup. There is no code path between the two, deliberately.
The list is used for one message: notice that API verification has launched. It is not a newsletter and is not shared with, sold to, or rented by anyone. Every email carries an unsubscribe link, and you can unsubscribe at any time without contacting us.
6. Server and security logs
The site runs on Cloudflare. Like every web server, the platform records standard request metadata — IP address, timestamp, user agent, requested path — for delivery, security and abuse prevention. We do not have access to your analysed text through those logs, because in local mode the text is never part of a request.
For rate limiting, our own server stores a salted, non-reversible HMAC of the client IP address together with an hourly counter. The raw address is never written to our database, and these counters are kept in a separate table from waitlist signups so the two cannot be joined.
Cloudflare Turnstile is used to keep automated traffic off the endpoints that cost money. It is a privacy-preserving challenge that does not track users across sites.
This site ships without analytics. If a privacy-friendly, cookie-free analytics tool is added later, this section will be updated before it goes live.
7. Retention and deletion
| Data | Retention | How to delete it |
|---|---|---|
| Text analysed in local mode | Never leaves your browser; discarded when the tab closes | Close or refresh the tab |
| Theme preference | Until you clear it | Clear site data in your browser |
| Waitlist signup | Until the launch notification is sent, or until you unsubscribe | Use the unsubscribe link, or email us |
| Account (email, balance, ledger) | Until you ask us to delete it | Email us and we will remove it |
| Sign-in codes | Minutes — single-use and short-lived, stored only as a hash | Expires on its own |
| Sessions | Until they expire or you sign out | Sign out, or clear cookies |
| Purchase records | As long as tax and accounting rules require | Cannot be deleted while that obligation lasts |
| Rate-limit counters | Pruned automatically; hashed, never the raw address | Expires on its own |
| Platform request logs | Per Cloudflare's own retention policy | Contact us and we will pass the request on |
To request deletion of anything we hold about you, or to ask what that is, use the contact route below. We will respond to a verified request within 30 days.
8. Children
The site is not directed at children under 13 and does not knowingly collect personal information from them.
9. Changes and contact
Any material change to this policy will appear here with a new “last updated” date. Sections currently marked as future previews will be rewritten as statements of current processing before the feature they describe is switched on, never after.
Questions about this policy can be sent to the address published in the site footer or on the repository that hosts this project.