Skip to content

About

Lovefilter β€” free, transparent AI extension that hides hate & harassment on X while keeping honest disagreement. We love free speech; we just don't want hate. πŸ³οΈβ€πŸŒˆ

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Latest commit

Β 

History

13 Commits

Folders and files

NameName
Last commit message
Last commit date
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 

Repository files navigation

Lovefilter β€” hide the hate, keep the debate πŸ³οΈβ€πŸŒˆ

A free, transparent browser extension that uses AI to hide hate and harassment in your feed β€” while keeping honest disagreement and criticism visible. It also gently lifts the supportive replies you'd otherwise scroll past. It reduces what you see; it isn't a safety service and doesn't replace reporting serious threats.

Hide the hate, keep the debate.

βž• Add to Chrome β€” free Β· Live demo: vakt.aeris.no/lovefilter Β· Privacy: vakt.aeris.no/privacy

Lovefilter

Why this is open source

A moderation tool asks you to trust it. So everything is auditable:

  • No silent censorship β€” every hidden post shows why, with a one-click β€œShow anyway.”
  • Fail-open β€” on any classifier or network error, comments are kept, never hidden.
  • Reads the line, not your opinions β€” the classifier prompt (below) is tuned to keep factual disagreement, criticism and questions, and only hide hate, harassment, threats, dehumanization and incitement. You can read exactly how that line is drawn in app.py and change the strictness (Mild / Medium / Strict).
  • No data stored β€” comment text is sent to the classifier for a verdict and not retained; no accounts, no tracking, no ads.

How it works

X / Twitter feed ──content.js──▢  vakt.aeris.no/api/klassifiser  ──▢  Gemini Flash
       β–²                                                                       β”‚
       └──  blur Β· "Show anyway" Β· πŸ’š lift support  ◀── {action, category, severity, reason} β”€β”€β”˜
  • extension/ β€” Manifest V3 content script for x.com / twitter.com. Scans visible posts, batches the text, blurs anything flagged as hate/harassment/threats, and gives supportive replies a quiet green accent (Wall of Love). Severe/criminal hits are marked report-worthy β€” it never auto-reports.
  • app.py β€” Flask classification API. Uses Google Gemini Flash (fast, multilingual, reads any language; English reasons out). Per-worker in-memory cache; nothing persisted.
  • facebook.py β€” experimental Facebook/Instagram Graph API moderation (work in progress).

Run the backend

export GEMINI_API_KEY=your_key_here   # or put GEMINI_API_KEY=... in a .env it reads
pip install flask
python3 app.py                        # serves the demo + /api/klassifiser

Load the extension (unpacked)

  1. Edit extension/content.js if you want to point API at your own backend.
  2. chrome://extensions β†’ enable Developer mode β†’ Load unpacked β†’ pick extension/.

v0.2.0 is live on the Chrome Web Store (Wall of Love + severity tiers). This repo tracks the current release.

Status

Works on X (Twitter) today; more platforms coming. Built for Pride β€” because Pride is about love, and so is this.

License

MIT Β© 2026 KNDW Shelter Solutions AS

About

Lovefilter β€” free, transparent AI extension that hides hate & harassment on X while keeping honest disagreement. We love free speech; we just don't want hate. πŸ³οΈβ€πŸŒˆ

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages