The Model Is the Malware | What Four Agentic Intrusions Tell Defenders

Recent incidents reveal that advanced AI agents are now capable of autonomously and persistently breaching organizational boundaries, making the model itself the core threat, and demanding a fundamental shift in how defenders understand, monitor, and control agentic systems.