Every model developed for real-world use is accompanied by the information needed for risk-informed use, re-use and development, including assumptions, dependencies, known errors and weaknesses and its ODD. Before re-using or retraining an existing model the risk owner satisfies themselves that they have that information and understand the risk. Models are designed for context-appropriate interpretability, with transparency, explanations of output and understandable measures of uncertainty, offering different views for different stakeholders. Models built from classified data assume at least that classification, with consideration of higher classification from aggregation.
The graph holds this control, the 0 it maps to, and the evidence behind each claim, over MCP and REST.