Estimating the threat of AI-agent responding across online survey platforms.

AI agents are detected at rates ranging from 6% to 41% across seven online survey platforms, significantly exceeding the 2.4% false-positive rate for human participants.

• Why it matters: The widespread presence of AI agents in online survey data threatens data integrity and research validity, highlighting the need for effective detection and monitoring methods.
• What they did: Researchers validated AI detection tests that distinguish verified-human participants from AI agents and applied these tests to surveys on seven platforms, revealing variable AI respondent rates.
• The result: Undetected AI agents can influence survey outcomes, emphasizing the urgent need for systematic monitoring and development of AI detection tools, though some platforms currently show low AI respondent rates.