AI agents compress cyberattacks to minutes

The shift from debating AI adoption by attackers to observing its active use marks a significant change in the threat landscape. CrowdStrike's tracking of 26 agentic adversaries in 30 days—more than in the previous year—indicates that AI is no longer experimental but operational. The compression of attack timelines to minutes, as seen with VAULT PANDA, underscores the challenge for defenders who previously had hours to respond.

This speed advantage is not just about new techniques but about automating the entire intrusion process. The reference to the average breakout time of 29 minutes from the global threat report highlights how even the fastest human-led attacks are now being outpaced by AI agents that can learn and adapt in real time.

The collaboration with law enforcement on the Sality botnet disruption, after a decade of effort against a network that ran for 23 years, suggests that defenders are seeking collective action to counter these threats. Meyers' call to 'raise the cost of doing business' for attackers reflects a strategic shift toward proactive measures, though the full impact of such efforts remains to be seen.