I feel like you'd suffer heavily from adverse selection.
That's a good point. This is where working with experienced actuaries to build risk models will be important. This idea would not be a tech startup per say. Just a good 'ol insurance policy with a shiny iOS app wrapper :)
1. The crime occurred in another city about 3 hours way, so poor granularity or precision really didn't come into play.
2. The crime itself was a hit and run on a road worker. Not something you'd generally believe to be premeditated where I'd have thought to have a friend haul my phone around town to provide an alibi.