AI Expands Security’s Control Plane
The week moved AI security from agent identity and automated patching to demonstrated exposure: Copilot could leak 2FA codes, while spyware tried to interfere with AI malware analysis. AI therefore emerged as both a governance layer and an attack surface.