AI-Driven Cyber Threats Surge with Agentic Adversaries

In September 2026, cybersecurity reports reveal a significant rise in agentic adversaries utilizing AI for intrusions, reducing attack timelines to mere minutes. CrowdStrike noted that the average breakout time for intrusions has dropped to 29 minutes, with some operations completed in under a minute. Concurrently, Beazley Security reported a 36% increase in newly disclosed vulnerabilities attributed to AI's influence on security research. Despite this increase, traditional attack methods, particularly credential theft, remain the primary vector for ransomware attacks, accounting for 67% of incidents. The Sality botnet disruption was highlighted as a major defensive effort against these evolving threats. The landscape is shifting rapidly, with AI both complicating defenses and enabling faster attacks.

Key Points: • Agentic adversaries are reducing intrusion times to minutes using AI. • A 36% increase in disclosed vulnerabilities was reported in Q2 2026. • Credential theft remains the dominant method for ransomware attacks.