For Kant all physics is an account of appearances. Classical mechanics, relativity, quantum field theory all of it belongs to epistemology (how we know), not ontology (what exist). None of it tells us what exists “in itself.” Kant would have had no trouble saying that black holes, electrons, entanglement are lawful appearances under conditions of possible experience. Full stop.
Now, physics constantly revises its laws because the same human apparatus encounters appearances under different constraints. Bus-scale laws fail near black holes. Classical probability fails at quantum scales. This is not surprising; it is exactly what Kant predicts once you stop mistaking local success for necessity.
Where non-locality does push you toward Kant is subtler than your current objection might allow. Non-locality does not challenge Kant’s claim that physics is epistemology. It challenges the physicists’ unexamined realism inside that epistemology.
Before quantum theory, physicists quietly assumed that appearances map onto a world made of spatially separate entities with intrinsic properties. That assumption was never justified by experience; it was smuggled in from common sense. Kant explicitly warned against this move by saying that spatial multiplicity does not license metaphysical claims about things as they are in themselves.
Bell non-locality exposes exactly that mistake. What fails is not “space as appearance,” but the inference that because two things appear spatially separated, they must be ontologically independent. That inference is pre-Kantian metaphysics masquerading as physics.
When experiments force physicists to give up local hidden variables, what collapses is not Kant’s framework but a naïve realism that Kant already dismantled. Non-locality shows that even within appearance, spatial separation cannot be read as metaphysical separability. That is a Kantian correction, not a refutation.