This guide covers what your authenticity check results mean and how to act on them. For what these checks do and when to use them, see What are authenticity checks?
Where to find your results
Go to your study's 'Participants' page in Prolific. Submissions are split into three tabs: 'Awaiting review', 'Approved', and 'Rejected'. Two columns show your results: 'Auth check: LLM' and 'Auth check: Bots'.
This page is the same regardless of which platform you built your study on, since the checks run on Prolific's side.
What your results mean
Each result falls into one of three bands:
High (green): no suspicious signals were found. No action needed.
Mixed (orange): some signals were found, but not consistently. Worth a closer look.
Low (red): most signals point to non-authentic behavior. Worth a closer look.
You might also see no result at all. This usually means the check hasn't finished running yet, or it doesn't apply to that particular submission. For example, an LLM check has nothing to assess if a participant didn't answer any free-text questions.
How to review a flagged submission
Look at the submission itself. Use 'Download demographic data' on the Participants page to review the participant's responses question by question, especially for a flagged LLM check result.
Consider the context. A mixed or low result isn't proof on its own, participants can have reasonable explanations. You can message participants directly using 'Message selected' before making a decision.
Decide whether to approve or reject the submission, using 'Approve selected' or 'Screened out' on the Participants page.
A note on quality assurance
Prolific also reviews mixed and low authenticity results as part of its own data quality audits.
Click here to learn more about Prolific’s work in data quality, and about our human data guarantee campaign.
What's next
How to set up authenticity checks: Qualtrics, Gorilla, Pavlovia, LimeSurvey, AI Task Builder