(From memory, so possibly miss some details) Differential Privacy uses the concept of adding 'noise' to a dataset to make it statistically provable, that, if q

(From memory, so possibly miss some details) Differential Privacy uses the concept of adding 'noise' to a dataset to make it statistically provable, that, if queries are supposed to only be A, B, C... then the adversary can only tell the target from noise with some probability - The point of restricting access to a subset of queries isn't really the main point of how DP improves privacy, but rather, restricting access to a subset of queries makes the formal proofs of things doable.