The provider must continuously monitor model inputs to guard against malicious input attacks such as injection, data theft and adversarial attacks, and must establish routine monitoring and evaluation together with model emergency management measures, promptly handling security problems found while the service is running and optimising the model through targeted instruction fine-tuning, reinforcement learning or similar methods.
The graph holds this control, the 0 it maps to, and the evidence behind each claim, over MCP and REST.