How it works
Gaze Keeper works on several levels, combined so it stays effective even when one of them isn’t enough on its own:
- Network blocking: before anything even starts loading,
the visited domain is checked against several combined block lists :
- the default block list, built into the extension.
- the dynamic block list, filled in automatically by the heuristic analysis (see below).
- the manual block list, domains you add yourself.
- imported block lists, one or more remote sources you enable from the settings page.
- Allow list: a list of trusted domains, built into the extension (search engines, code-hosting platforms…), can never be added to any block list, whichever one it is.
- Blocking a page through content analysis: the page’s visible text is analysed locally by Gaze Keeper (nothing is sent out) and matched against built-in dictionaries of terms commonly used on pornographic sites. If several explicit terms are found, a full-screen veil showing a Bible verse replaces the page’s content; the site then becomes a candidate for the dynamic block list, and is added to it after several blocks. This block applies even on a site that’s on the allow list. This veil resists a quick removal attempt (developer tools) — it’s not a security lock, just resistance to impulsive removal.
- Blocking media: an image or video whose associated text (caption, description, surrounding content) is explicit is neutralised individually, even if the rest of the page isn’t otherwise blocked. This neutralisation also applies on a site that’s on the allow list.
Heuristic analysis (URL and page content) relies on dictionaries covering around twenty languages (French, English, Spanish, Portuguese, German, Russian, Italian, Japanese, Arabic, Chinese…), loaded based on your browser’s language and the visited page’s language — even though Gaze Keeper’s interface is only available in French, English and Spanish, detection also works on sites in other languages.
Still unclear? Suggest an improvement to the docs.