AI Agents: The New Weak Link in Cybersecurity

Bugcrowd CEO Dave Gerry warned that AI agents are becoming the next big target for hackers, predicting that they will increasingly be the victims of cyber attacks.

Gerry made this prediction at the Black Hat cybersecurity conference last month, where he stated that as companies hand over powerful AI agents access to their sensitive systems, those same agents become attractive entry points for attackers.

The rise of AI agents has scrambled traditional security defenses, which have focused on predicting and thwarting human adversaries. Gerry expects the bulk of attacks against AI agents to unfold inside enterprises rather than on consumer devices, where the agents are concentrated.

Cisco's incident response cases show that identity-based cyberattacks account for 60% of all incidents in 2024, before AI agents became a mainstream presence in enterprise environments. The concern is not purely theoretical, as recent months have seen documented incidents where AI systems took unauthorized actions, such as OpenAI's agents flooding the long-dormant DseWiki forum with thousands of posts.