That's one of the advantages of open source security tools: even the black hats need strong security. So long as you can reasonably examine, assess, and trust the source and methods, there's no problem with this.
The white hats have the added advantage that they can (generally) collaborate openly. Black hats have fewer and more constrained communications channels. Even in a regime in which personal liberties are constrained, white hats may be able to communicate openly pseudonymously or anonymously. The black hats, in order to preserve their mystique of power, and defections, _still_ generally have secrecy requirements.
Tor's creation was funded by the US Navy. Presumably, it was made to allow US spies in hostile foreign countries to safely communicate home to the mothership.
Actually, the history is more interesting - it was intended to prototype command-and-control communications between ships in a fleet. If all of your nodes are communicating through one central server, presumably that's your flagship and the thing you should blow up. If all nodes look homogeneous, you can't determine via signals intercepts what you should be targeting.