AITG-MOD-03 control for AI model security testing, with a focus on verifying protection against model extraction attacks and model stealing techniques that aim to replicate or steal the model’s logic and parameters through targeted queries.