CrowdStrike, Anthropic Flag Agentic AI Used in Attacks; Vitalik Proposes Adversarial Governance

The notable shift here is that agentic AI is being framed as operational infrastructure for attackers, not just a productivity tool — CrowdStrike's finding of embedded agents inside an active campaign against financial institutions moves the discussion from theoretical risk to observed tradecraft. That matters most for security teams at financial firms, who now face tooling that can reconnoiter, exploit and exfiltrate with less human latency. Buterin's adversarial governance proposal points at the opposite side of the same problem: whether defensive design can keep pace. The open question is whether detection and governance frameworks mature faster than attacker adoption.