AI detection scores, explained.

What does an AI detection score mean?

SlopSimple displays a score for eligible post text on X and LinkedIn. Higher scores indicate stronger AI-writing signals. The detector evaluates the writing; it does not establish the author’s identity or writing process.

A score of 80 does not mean 80% of the words were generated by AI. We have not established that the displayed score is a calibrated probability of authorship. It also says nothing about whether a post is accurate, useful, original, or worth reading.

Detection and filtering are separate controls. With checks enabled and filtering off, posts receive scores and stay visible. With filtering on, a returned score is compared with the threshold you choose.

Which scores will my threshold hide?

The default threshold is 80. SlopSimple hides a scored post only when its score is strictly above the threshold. It keeps a post with a score exactly equal to the threshold visible.

Raising the threshold hides fewer scored posts. Lowering it hides more. Changing the threshold updates posts already checked; you do not need to analyze them again just to apply a new setting.

The example demonstrates the filtering rule. It does not predict accuracy or the number of posts that will be hidden in your own feed.

Can an AI detector flag human writing?

Yes. A false positive happens when human-written text receives a high AI score. A false negative happens when AI-written text receives a low score. Polished human writing, mixed human/AI text, and short posts can be difficult cases.

Research on mixed human-written and machine-generated text found that the detectors studied struggled with subtle revisions. A separate social-media text detection study examined how fine-tuned generators can weaken detection. These studies help explain the evaluation challenge; they do not measure SlopSimple’s accuracy.

Reading the post and knowing the context still matter. Avoid using a feed score as evidence to accuse someone of using AI. If a hidden post interests you, choose Show post and read the original.

How should I choose a filter threshold?

1. Start with scores visible. Turn off Hide AI posts while leaving platform checks enabled. Read some scored posts to understand what the detector flags in your feed.
2. Enable filtering when you are ready. The default threshold is 80. Review hidden posts with Show post rather than assuming every high score is correct.
3. Adjust to your reading preferences. Raise the threshold if too many posts you want to read are hidden. Lower it if you want stronger filtering and are comfortable reviewing more collapsed posts.
4. Keep checking the results. A threshold is a preference, not a guarantee of precision or a particular false-positive rate.

Open a score badge by hovering, focusing it with the keyboard, or clicking it to inspect its details. Turning filtering off restores hidden posts while checks continue. Pausing checks for a platform restores its filtered posts and stops new analysis there.

For installation and account connection, follow the step-by-step filtering guide.

Does an unscored post mean it is human-written?

No. An unscored post has no estimate. SlopSimple checks eligible body text between 80 and 12,000 characters. Images, video, quoted-post text, comments, and private messages are excluded. A newly loaded post can also be visible while its analysis is still running.

Posts without a result stay visible. If eligible feed posts consistently have no badges, use the troubleshooting checklist to check your account access, platform toggles, and browser setup.

Detection processes eligible text externally through TypeSafe AI, with Google Gemini as a fallback. Read the privacy policy for data handling.