I’ve only ever done QFT in curved spacetime, but I don’t see any reason why you couldn’t do EM, it’ll be a vaguely similar process. I never actually dealt with any scenarios where the curvature was that extreme, and QFT in a curved background is kinda bizarre and doesn’t always require one to consider the specific trajectories, though you definitely can especially if you’re doing some quantum teleportation stuff. In my area it’s simpler to ignore QED and to just consider a massless scalar field, this gives you plenty of information about what photons do without worrying about polarisations and electrons.

It’s been a long time since I did any reading on the geometric optics approximation (in the context of GR this is the formal name for light travelling on null geodesics), but for the most part it’s not something you have to consider, even outside of black holes the curvature tends to be pretty tame (that’s why you can comfortably fall into one in sci-fi), so unfortunately I don’t know of any phenomena (in GR) where it’s important. QFT in curved spacetime generally requires you to stay away from large curvatures, otherwise you start entering into the territory of quantum gravity for which there is no accepted theory.

Outside of GR, it breaks down quite regularly, including I believe, for the classic double slit experiment.