AI Slop Detector for YouTube автор Jason Tiller

AI Slop Detector rates YouTube videos for signs of AI-generated content — TTS voiceovers, stock footage, clickbait titling — before you click. It runs entirely in your browser using public metadata only; no account, no server, no tracking.

There's no audio or video analysis (that's not something a browser extension can cheaply do in real time). Instead, it weighs several public signals against each other: YouTube's own "altered or synthetic content" disclosure label when present, how often a channel uploads, a channel's age versus its output volume, clickbait/content-mill title patterns, AI-tool boilerplate or stock-footage credits in the description, and whether a channel's recent uploads all follow the same templated title.

Wherever you're watching, you'll see one of four plain-English labels — Likely AI-Generated, Likely AI-Assisted, Mixed/Uncertain, or Likely Human-Made — with a "Why?" section listing the actual reasons behind it, not just a bare score. A small badge does the same on thumbnails across your feed and search results.

Disagree with a call? Mark any individual video as human-made or AI slop yourself — your override always wins over the heuristic, and only affects that one video. Settings let you adjust sensitivity, toggle each surface on or off, and optionally add your own YouTube Data API key for a richer channel-history check.

Take it as a strong hint, not a verdict — this is a heuristic aid for your own judgment, not a fact-checker.