Long2Text
← Long2Text

Privacy policy

Last updated 2026-09-28

Your screenshots

Your screenshot is sent through our server to Google Cloud Vision for text recognition. Large images may be split in your browser. Tencent Cloud OCR is available as an alternative provider when configured. Long2Text does not store your images; they are processed in memory and discarded after recognition.

AI proofreading

When AI proofread is on, a small number of lines the OCR engine was unsure about — together with a few nearby lines of text — are sent to SiliconFlow's model API to fix misreads. You can switch it off at any time; the setting is remembered in your browser.

Pictures found in your screenshot (stickers, photos) are cut out in your browser. Only if you click “Describe them with AI” are those small crops sent to SiliconFlow's vision model to get a one-line description; the full screenshot is never sent.

Usage limits

To keep the free converter from being abused, we count how many recognition, proofreading and picture-description requests each network address makes per hour. The address is stored only as a keyed one-way hash, never in readable form, together with the hour and the count, and is deleted within two days.

If you sign in

Google sign-in gives us your name, email address and profile picture, which we use only to link your purchases and history to you.

Recognized text from images you convert while signed in is saved to our database so you can find it again in History.

If you use the API

API keys are stored only as a one-way fingerprint (SHA-256). The full key is shown to you once, when you create it.

For every API request we keep a usage record: the time, the key, the result, the image's file size and dimensions, the number of characters and pictures, and how long it took. We use it to charge credits and to show your usage in the dashboard. We don't keep the image or the text read from it.

Screenshots sent to the API go through the same recognition and proofreading as on the website. In the `text` and `both` modes, the pictures cut out of the screenshot — never the whole screenshot — are sent to SiliconFlow's vision model for one-line descriptions.

Payments

Payments, including API credits, are processed by Stripe. We receive a confirmation of the payment, never your card details.

Analytics and cookies

Optional Google Analytics is off by default. You can enable it using the preference below and disable it here later. Do Not Track, Global Privacy Control and an existing refusal keep it off. We do not load Google Analytics before your opt-in.

When enabled, it measures public-page visits, visible pricing/paywalls, sign-in attempts and successful returns, conversion outcomes and duration ranges, input methods, preview versus full exports, checkout attempts and verified payment returns. It does not receive screenshot content, file names, account or payment identifiers, or raw error messages. These statistics describe consenting visitors only and are not our payment ledger.

We discard raw URL queries and full referring URLs. Only predefined source labels (such as Reddit or Discord) and campaign labels are accepted. After opt-in, these labels can persist in this tab for up to 30 minutes, and an identifier-free login-attempt marker for up to 15 minutes. Revoking analytics clears these markers. Attribution starts after consent; it cannot recover earlier visits.

Necessary cookies support sign-in and language choice. Optional analytics cookies are used only after you enable analytics; your preference is stored locally in this browser.

Optional analytics preference

Off by default. If you enable it, Google Analytics uses cookies to measure visits, conversions, exports and payment confirmations. It does not receive screenshot content, file names or payment identifiers. All features work with analytics off.

Current status: Off

Saved in this browser only. Changing this preference reloads this page.

Questions or deletion requests: support@long2text.com