From Shadow IT to GhostOps: The Rise of Unauthorized AI Agents in the Enterprise

**GhostOps** describes the uncontrolled spread of unauthorized operational AI agents within enterprises and the elevated risks they create, including data residency breaches, credential/secret sprawl, privilege escalation via automation, prompt injection and tool abuse, open-source supply-chain compromise, loss of auditability, regulatory/IP exposure, and runaway cost; illustrated through recurring patterns (developer productivity bots, finance reporting agents, helpdesk autopilot). The report recommends a two-track approach—governance by design (agent inventory, reference architectures, identity/secrets hardening, data protection, model/agent risk management) and detection by evidence via SIEM/MXDR (e.g., Microsoft Sentinel)—to enable safe, rapid adoption with enterprise-grade visibility and control.