Image Text Selector — Privacy Policy

Effective: 12/06/26. Applies to the Firefox extension
"Image Text Selector" and its companion macOS helper program, version 1.0.0 and later.

The short version: this extension collects nothing, transmits nothing to us or to
any third party, and contains no analytics. All text recognition happens on your own Mac.

What the extension does with images

When you hover over a large image, the extension downloads that image's bytes from
the same website that is already displaying it. That request is sent without your
cookies and without telling the site which page you were on — it is simply a
second request for a picture the site already showed you. The bytes are handed to a
small helper program on your Mac. The helper uses Apple's built-in Vision framework
to recognize the text, entirely on your device, and returns the recognized words
to the extension so you can select and copy them. The image and its text never leave
your computer.

What we store

• One on/off setting is saved in Firefox's local extension storage.
• Recognized text is kept briefly in memory so re-hovering an image is instant:
  at most about 2 MB in the extension's background process (erased automatically
  within about half a minute of inactivity, and whenever Firefox closes) and at most
  about 512 KB per tab (erased when you close or leave the tab, or turn the
  extension off).
• Apart from that single on/off setting, the extension itself writes nothing to
  disk: recognized text lives only in memory, and image pixels are never retained
  by the extension at all. The image request participates in Firefox's normal web
  cache exactly like the page's own image load did (and bypasses that cache entirely
  in private windows).

What we collect

Nothing. No telemetry, no usage statistics, no crash reports, no identifiers, no
accounts. We have no servers. The extension's Firefox data-collection declaration is
"none", and Firefox shows this at install time.

Private browsing

The extension only runs in private windows if you explicitly allow it in Firefox
(about:addons → this extension → "Run in Private Windows"). If you do, images
hovered in private windows are fetched without using or writing the browser's disk
cache, and their recognized text is not kept in the shared cache.

The helper program

The companion helper is a small open-source program you install yourself; its full
source code is one Swift file in the public repository (https://github.com/fffoivos/image-text-selector). It has no
network capability: it reads image bytes from Firefox through a local pipe, runs
Apple's on-device text recognition, writes the result back, and exits. It creates no
files, keeps no logs of your images or text, and no copy of it stays running between
uses. Apple's Vision framework processes images on-device and does not send them to
Apple.

Clipboard

Text is placed on your clipboard only when you yourself copy a selection (for
example with Cmd+C). The extension never reads or writes the clipboard otherwise.

Changes

If a future version ever changes any of the above, the change will be stated in the
version notes and reflected in Firefox's data-collection disclosure before it ships.

Contact

Questions or concerns: [CONTACT — recommended: the GitHub Issues page of REPO_URL,
which avoids publishing a personal email address].