Just the knowledge that something CAN be used incorrectly, for dangerous purposes, is often the same as the knowledge that something WILL be used incorrectly, or for dangerous purposes, when dealing with large groups of people. A large enough sample set almost guarantees that SOMEone out there is up to no good.
This isn’t necessarily true on an individual basis, though, and that’s where most security profiling falls flat. In terms of justification of action, the mandate for intervention is lost when dealing with a presumably innocent (as must legally be presumed) individual.
The number of people that must be involved in an exploitable system before things turn sour must always be examined before placing into action preventative measures that will inconvenience everyone. If the number of people is high enough, nothing should be done, so as not to unduly punish a large group of people for the potential crimes of the few.