Assuming the validity of the equivalence principle in the quantum regime, we argue that one of the assumptions of the usual definition of quantum mechanics, namely separation between the “classical” detector and the “quantum” system, must be relaxed. We argue, therefore, that if both the equivalence principle and quantum mechanics continue to survive experimental tests, then this favors “informational” interpretations of quantum mechanics (where formalism is built around relations between observables, defined as information that can be accessed by an observer of the system. In particular “collapse” is understood as a change of relative information as the detector interacts with the system) over “ontic ones” (assuming the physical reality of states and wave functions, which are assumed to be more than informational objects. In particular collapse is understood as a physical process). In particular, we show that relational-type interpretations can readily accommodate the equivalence principle via a minor modification of the assumptions used to justify the formalism. We qualitatively speculate what a full generally covariant quantum dynamics could look like, and comment on experimental investigations.