AI-agent orchestration is the coordination of multiple automated agents, tools, or workflows to achieve a shared goal. An orchestration system may assign tasks, pass information between agents, call external services, and manage the sequence of actions.
In cybersecurity, orchestration can concentrate access and authority: an agent may act through service accounts, API tokens, or administrative tools. If an attacker gains control of an account or abuses legitimate workflows, they may be able to automate actions across connected systems. Defenders should restrict management interfaces, apply least privilege, protect credentials, and monitor task creation, administrative changes, and outbound activity. Orchestration itself is not a vulnerability; the risk depends on its configuration, permissions, and how access is used.