Hides insults, slurs, threats and swearing on web pages. Optional on-device AI covers 14 languages. Nothing leaves your device.
Users
0
store count
Rating
0.0 / 5
unchanged in 30 days
Ratings
0
total ratings
Store rank
#367K
#301K among extensions
Version
1.0.1
updated 19 Nov 2026
Publisher and listing changes
From our daily record of this listing since 18 Nov 2026
No change of publisher name, contact email, website, privacy policy or listing name since 18 Nov 2026.
About Detox
Detox hides toxic comments on the pages you read: insults, slurs, threats and strong swearing. Everything happens on your computer; Detox never sends the text you read anywhere.
HOW IT WORKS
• Toxic comments are blurred. Click one to read it anyway.
• Or mask just the bad words with •••.
• Three levels: Relaxed (slurs and threats), Balanced (also insults and strong swearing) and Strict (also mild rudeness like "stupid").
• Catches disguised spellings such as 1d10t, f*ck and a$$hole, while leaving words like "Scunthorpe" and "assessment" alone.
• Works on comments that load as you scroll, and on tabs that were already open when you install it.
• The toolbar badge shows how many items were hidden on the page.
OPTIONAL AI CHECK (OFF BY DEFAULT)
The built-in word list knows English words. Turn on the AI check and a toxicity model also reads whole comments, catching insults in 14 languages: English, Russian, Ukrainian, German, Spanish, French, Italian, Japanese, Chinese, Arabic, Hindi, Hebrew, Tatar and Amharic. It is strongest in English, Russian and French.
The model runs on your computer. It is downloaded once from Hugging Face (130 MB, or a 28 MB lite version), and you can delete it from the settings page at any time.
YOUR CONTROLS
• Switch Detox off everywhere or on one site.
• Add your own words to always hide, and words to never hide (such as a name).
PRIVATE BY DESIGN
No account, no server, no analytics, no ads. Detox doesn't collect, send or sell anything. Source code: https://github.com/searope/detox
LIMITS
No filter is perfect. A word list can't read intent, and the AI check misses some hostile remarks written without insulting words, and can flag quotes or discussions about slurs. Use the "never hide" list for false alarms.
Version history
2 versions recorded
No update since 18 Nov 2026 changed the permissions or site access.