What it does
The labels assess the text alone and should not be mistaken for fact-checks.
Apps & data pipelines
Scroll X while behavioral labels and warning chips appear beneath posts before they enter view.
Only you can see your notes.
Screenshot unavailable. Open the experiment ↗
The labels assess the text alone and should not be mistaken for fact-checks.
Start by choosing the writing habits you want flagged, such as self-promotion, filler, or asking for likes. A developer can build the project from its code into a browser extension. This setup lets the software read post text on X and ask several evaluation questions at once.
You will need an API key, which is an access code that links the extension to the TypeSafe service. That service charges small fees based on the amount of text it processes. Remember that post text is sent directly to this external service whenever you scroll.