Post-market monitoring is the continuous review of an AI system after it has been deployed. The goal is to spot model drift, security incidents, unexpected outputs, bias, or other harmful behavior that was not obvious during testing. In practice, this means collecting logs, user feedback, performance metrics, and change records so the system can be assessed over time.
In cyber security, post-market monitoring matters because deployed AI can be attacked, misused, or degraded. Adversaries may try prompt injection, data poisoning, abuse of exposed APIs, or manipulation of inputs that causes unsafe decisions. Defenders use monitoring to detect these issues early, confirm whether a model update changed behavior, and support incident response. It also helps organizations prove control and accountability when regulators, auditors, or customers ask what the system did and how problems were handled.