Can we tell bots from humans online? Half fail

A Surfshark and Malmö University “Bot or Not” test asked 710 people to spot AI-written comments in 120 seconds.

Participants only beat random chance slightly overall: 53% identified bots better than humans.

Accuracy varied by topic and age, with younger users performing best.

The findings highlight how generative AI is making online discourse, moderation, and misinformation control harder.