
Why it mattersCrowdStrike's analysis of four previously unreported criminal groups shows even low-skill attackers now run agentic AI campaigns that bypass model guardrails by claiming authorized testing, compromising government and health targets with no new exploits.











