AI is now creating more defender workload than attacker handwork. The practical problem is not abstract use of AI by adversaries; it is that AI can generate enough believable activity to bury real signals in the queue and speed exploit development beyond normal response cycles.
CrowdStrike says AI-driven malicious activity rose 89% year over year, AI-triggered detections outnumber human-triggered ones 2.5 to 1, and its systems triaged an average of 14 million detection leads a day to produce about 36,000 customer alerts. The report also says 88% of vulnerabilities were weaponized through AI within 48 hours, which makes the old patch window too slow for the pace of AI-assisted abuse.