The extension
The card explained: verdict, confidence, detectors and skipped checks
What each row of the card means, what the badges mean, and why some checks say they were skipped.
Updated
Tap the circle and a white card opens, anchored above it. It uses the platform's own system font so it reads as part of the page.
Header
The squash mark, "Slop Squash", and on the right what was checked, for example "Post · 3 images".
Verdict row
The headline is one of AI-generated (a provenance signal fired, or the score is at least 95%), Likely AI-generated, Uncertain, or Looks human. On the right is the confidence as a percentage, with a ten-segment bar under it. Confidence is the combined score of every detector that ran; see How verdicts are scored.
Detector rows
One row per detector that produced a result, each with a title, a one-line finding and a badge:
| Badge | Meaning |
|---|---|
| Strong | A provenance signal that asserts the content is AI-generated: a trusted Content Credentials signature, a matching watermark check, or metadata naming a generator. |
| Likely | A probabilistic detector (writing pattern, generator classifier, model recheck) scored the content as probably AI. |
| Agrees | Community reports lean the same way as the verdict. |
| Checked | The detector ran and did not find a signal. |
Row titles you will see: Image watermark or Image provenance, Content Credentials, Creator metadata, Text pattern, Generator classifier, Model recheck, and Community.
The skipped line
Under the rows sits one muted line such as "Checked the video's thumbnail · 4 checks skipped · Show". Show expands it into every detector that could not run and why, and every detector that ran without an opinion and what that means. Examples: "SynthID text watermarks are keyed; not checkable", "Video frames not analysed; container and poster only", "Animated image: provenance only". Whether the list is open is remembered for the browser session. The card always lists every detector, so you know what was and was not examined.
Footer
Hide post (hides that post for the session), Not AI (a community report), and, when the verdict is Looks human, Report as AI. Pro adds Always hide this account, Recheck with model, Always allow this account and Always allow this post. On the free tier those controls are simply absent, not greyed out. See Free versus Pro controls.