Prolific's authenticity checks help you collect genuine human data by flagging responses or behavior that may not reflect authentic human participation.
There are two types of checks. They address different risks, so we show the results separately, and you should read them separately, too.
LLM authenticity checks flag when a participant may have used an LLM (like ChatGPT) to answer free-text questions. They look for suspicious behavior, like copying and pasting or switching tabs, on free-text questions only.
Bot authenticity checks flag when an AI agent or automated bot may have completed your study. They look for non-human or scripted behavior across all question types.
When you opt in to authenticity checks, Prolific doesn't view your submissions. The process is fully automated: once it's run, the only data we can access is the result (the likelihood of non-original content or non-human behavior).
Which platforms are supported?
Authenticity checks are available on Qualtrics, Gorilla, LimeSurvey, Pavlovia, and AI Task Builder.
Check type | Supported on |
LLM checks | Qualtrics, LimeSurvey, Pavlovia, and AI Task Builder |
Bot checks | Qualtrics, LimeSurvey, and Pavlovia |
A few things to note:
On AI Task Builder, LLM checks are on by default for every study. They're only meaningful if your study meets the criteria below.
Authenticity checks aren't available for Taskflow studies.
LLM authenticity checks
These checks look for behavioral patterns that suggest a free-text response wasn't authentically human-written, like a participant copying and pasting an answer from ChatGPT or another LLM. They analyze behavior, not the words themselves.
Use them when your study:
Has free-text questions
Needs participants' own thoughts, opinions, or experiences
Uses one question per page (so we can link behavior to a specific question)
Don't use them when your study:
Has no free-text responses
Needs only very short answers (inauthentic behavior is harder to detect)
Asks participants to research information externally
Asks participants to summarize or reference external sources
Needs tools or resources outside the study
For example, avoid a prompt like this, since it requires participants to research and summarize external material:
"Visit Wikipedia and research the history of coffee cultivation. Write a 150-word summary of how coffee production spread globally."
Bot authenticity checks
These checks look for interaction patterns and behavioral signals that suggest a bot, agent, or script completed your study, rather than a human. They focus on behavior, not the quality of written responses.
Use them when your study:
Is vulnerable to automation or scripted behavior
Depends on behavioral interaction patterns
You can run bot checks alongside LLM checks, since they cover different risks.
Don't use them when your study:
Requires participants to use AI tools (including LLMs)
Requires a virtual machine
Requires accessibility tools
Must be completed on a specific device, like a tablet
Tips for better results
If your study needs original, human-written answers, say so directly in your instructions. Participants who understand why authenticity matters are less likely to look for shortcuts.
A few things that help:
State clearly that AI tools aren't allowed, if that's a requirement of your study
Briefly explain why you need participants' own words, rather than just asking for them
Use one free-text question per page, so a result can be linked to the exact question it applies to
For example:
"Please share your personal experience with social media and how it has impacted your daily life. Write a thoughtful response of at least 150 words. Do not use AI tools or external sources. We are interested in your genuine personal experiences."