Edit: Ok, let me not be so brief. In short, the current detectors for monitoring reactors are either HUGE (rooms and rooms full of highly purified water, and even those just look at nearby reactors and only get 1 count every 3 days). Or the detectors are very close to the reactor (7 meters i.e. https://www.iaea.org/safeguards/symposium/2010/Documents/Pap...). So signal to noise is the main issue first of all.