Enforcement control is a technical rule or mechanism that actively stops, blocks, or limits an action when it would violate policy. Unlike a warning or audit log, it has authority to change the outcome: deny a request, cap a process, terminate a session, or quarantine a system.
In cyber security, enforcement controls turn security policy into something the system can actually obey. They appear in firewalls, access control systems, rate limiters, sandboxing, and runtime watchdogs that can interrupt unsafe behavior. This matters because attackers often try to push software past its intended boundaries through abuse of privileges, prompt injection, tool misuse, or automation loops. A strong enforcement control reduces blast radius by making the final decision external to the vulnerable component. In AI agents, for example, an out-of-band control can stop tool calls or shut down execution when policy limits are exceeded, rather than trusting the agent to self-police.