TLDR Tech / InfoSec
1.5 million AI agents are at risk of going rogue
A survey of 750 IT executives found that 53% of the estimated three million AI agents deployed across US and UK enterprises are not actively monitored or secured. 88% of respondents reported suspected agent-related security or data privacy incidents in the past 12 months. Researchers warned that overprivileged agents operating with broad credentials and persistent access represent a new insider threat vector, with many organizations unable to even inventory how many agents they have running. Security teams should implement continuous runtime oversight, tiered access controls for AI agents, and agent discovery capabilities to address governance gaps outpacing deployment velocity.
Read the full story