I think it's pretty common knowledge that the Government is investing a lot of money into being able to do just that -- that should be enough.
I would think that doing something like this is less about the technology available, but thinking up an intelligent way of doing it. If you imagine a world where you have all the data at your finger tips: let's say you obtained every camera's footage, every phone's phonecalls,etc for a mile radius for a month before,during and after some disaster. Shouldn't be too hard to imagine.
The problem is: what are you looking for? How do each of the moments that you detect link to each other? What patterns are indicative of someone who is plotting something?
Those are very difficult questions, and that's what makes the problem difficult. Tackling a problem like that requires knowledge of voice recognition, NLP (Natural Language Processing), ML (Machine Learning -- If you don't want to put in every possible scenario yourself), and much much more. The rate of data creation (and thus data processing) would also be very high.
How to do this well is the difficult part.
If you still don't believe me: boil it down to just one source. Phones. Lets say you had a wiretap on every home in which a hitman was hired to (and succeeded in)killing a neighbor. It might be a good thing to run a simple text search on the word "kill", or "hit", and you'll catch the low hanging fruit (by this, I mean the people who are bad at hiring hitmen/commit crime), but it becomes more complex when you get more serious about catching people who aren't stupid enough to say exactly what they want to do, then you have to find patterns, assuming they are there.