slopstop

About

slopstop highlights posts and comments that read like they were written by AI.

As you scroll LinkedIn and Substack, slopstop scores the text of each post or comment against a set of heuristics — formulaic phrasing, em-dash density, hedge words, "It's not just X, it's Y" patterns, and similar tells — and shows a small colored badge inline with the content:

  • Green (0–29): likely human
  • Yellow (30–59): mixed signals
  • Red (60+): likely AI-generated

Hover any badge to see exactly which signals fired and how it was scored.

Where it works
  • LinkedIn — feed posts, reposts, and comments
  • Substack — full posts on /p/ pages, comments, and Notes

What it does NOT do
  • No data ever leaves your browser. There's no server, no telemetry, no account.
  • All scoring runs locally as you scroll.
  • The only thing stored is your on/off toggle (chrome.storage.sync).

Honest disclaimer
Heuristics aren't proof. Polished, formulaic human writers will trip false positives, and a careful prompter can write below the threshold. Treat the badge as a smell test, not a verdict — that's why every score links to its reasons.

Permissions
  • storage — to remember whether the extension is on or off
  • activeTab — so the popup can read the score totals from the current tab

Open to feedback. If a badge looks wrong, the reasons shown on hover will tell you why — that's usually the fastest way to suggest a heuristic to add or relax.