New research reveals that internet users fail to detect 60% of artificial intelligence bots on social media, with friendly and agreeable accounts proving significantly harder to spot than aggressive trolls.

A study by analyzed 1,722 participants worldwide through a live “Bot or Not” simulation. The findings show that while participants successfully flagged 50.2% of negative, confrontational AI-generated bots, their detection rate plummeted to just 38% when bots adopted a positive, friendly, or logical persona.

According to Luís Costa, Research and Insights Lead at Surfshark, polite bots pose a greater risk for social engineering attacks because they bypass human scepticism. Rather than provoking arguments, these accounts often validate users’ opinions to inflate discussion volumes or subtly spread disinformation without appearing threatening.

The simulation also highlighted how specific communication styles affect detection. Excessive emoji use served as a major giveaway, with emoji-heavy bots caught over 60% of the time, compared to a mere 35% detection rate for those keeping communication simple.

Furthermore, user demographics played a role. X users proved to be the strongest bot hunters with a 49% detection rate, significantly outperforming users on visual- and video-first platforms such as TikTok (38%) and Facebook (39%), leaving users on those networks more vulnerable to manipulation.

With automated detection tools and humans alike struggling to distinguish advanced AI content from real people, Surfshark recommends judging the account profile rather than the content itself. Users should inspect profile creation dates, generic photos, and bio details, while remaining cautious of suspiciously rapid support.

For more security expert tips on how to tell a bot from a human on social media, watch the video here:

For latest tech stories go to TechDigest.tv