Beyond conventional security threats, AI systems face threats aimed at the learning process, the model and its inputs, which the following sub-sections survey.
The graph holds this control, the 0 it maps to, and the evidence behind each claim, over MCP and REST.