Preferred
Chrome Web Store
Install the extension from the official Chrome Web Store listing.
Install from Chrome Web StoreInstall
Install from the browser store, or use manual install if you need it.
Chrome and Firefox do not generally allow true one-click third-party extension installs from arbitrary websites. The normal path is the browser store. Manual install is still useful for local testing, self-hosted distribution, and pre-store builds.
Also available
Firefox Add-ons
Install the extension from the official Firefox Add-ons listing.
Open Firefox listingManual
GitHub Releases
Download packaged releases and follow the manual install instructions for local testing or self-hosted distribution.
Open releasesManual install notes
For Chromium-based browsers, manual installation usually means loading the unpacked extension from developer mode. A ZIP download is useful as a release artifact, but it is not the same as a trusted one-click browser install.
How it works
Simple flow, visible signal, no account required.
Flag a video
If a video looks AI-made, a user can create the first record directly from the YouTube page.
Community votes
Other users vote up or down. The score is point-based, so each vote immediately changes the signal.
Confidence updates
The extension highlights low, medium, high, disputed, unknown, or unflagged states directly on YouTube.
Block if you want
Users who prefer a cleaner feed can enable blocking and temporarily reveal videos when needed.
Features
Built for real browsing, not a lab demo.
Inline YouTube indicators
Confidence badges, scores, and action buttons appear right where people are already looking.
Anonymous by design
No accounts are required. Votes are tracked through a local anonymous device ID.
Point-based moderation
Upvotes add one point, downvotes subtract one point, and heavily rejected videos can become unflagged.
Custom API support
The extension defaults to the project API, but advanced users can point it to another compatible backend.
Open source backend
The Fastify API and Prisma/PostgreSQL stack are public, inspectable, and ready for self-hosting.
Optional blocking
Highlighting is the baseline. Blocking is user-controlled and can be switched on only if desired.
Why this matters
AI spam scales faster than manual reporting.
The extension is meant to create a practical middle layer between “ignore it” and “wait for the platform.” Users get a visible signal, the community gets a shared database, and the project stays transparent because the code is public.
Open source
GitHub repository and API code available
Community driven
Detection improves as more people vote
Privacy-aware
No mandatory sign-in or public profile system
What users should know
Clear expectations before installing.
What gets stored
The extension stores settings and one anonymous device ID locally so one browser can only vote once per video.
What “unknown” means
Unknown is not a clean bill of health. It only means the community has not created a record yet.
How confidence works
Confidence is based on score ranges, not machine learning. It is an explicit community signal.
Who should use it
Anyone who wants a more transparent way to spot synthetic, AI-heavy, or spammy videos in their feed.
FAQ
Questions people will ask before they trust it.
Is this official YouTube moderation?
No. It is an independent community moderation layer delivered through a browser extension.
Does it use AI to detect AI videos?
Not in the current version. The detection signal is community voting backed by a shared API and video database.
Can I inspect the code?
Yes. The extension and the API are open source and linked directly to the public GitHub repository.
Can I run my own backend?
Yes. The extension supports changing the API server URL, so compatible self-hosted servers are possible.
Source and project status
Follow development, inspect the code, or contribute.
The project includes a Fastify API, a PostgreSQL-backed video database, and a Chromium extension that integrates directly into YouTube. If you want to review the source, open issues, or contribute, start on GitHub.