(This question is similar in spirit to this question, but I think this case is sufficiently different and specific enough to warrant a new question.)
In one of my current side projects, I have a little procedural name generation system for NPCs. Its purpose is akin to the random profiler in Watch_Dogs and the way it procedurally generates little NPC identities and blurbs to flesh out the game world, in that the system exists to spit out some generic NPC identities and give bits of information about them. Which can sometimes give very funny results, in Watch_Dogs’s case.
The procedural system in my project is fairly basic, and it generates:
- a random first name, taken from a list of the most common baby names in North America. (i.e. Anna, Nancy George, William, etc.)
- a random last name, taken from a list of the most common surnames in North America. (i.e. Rogers, Smith, Walker, Brown, etc.)
As a result of this dataset, the generator typically spits out generic names like "John Smith" and "Marcy Walker," which is what I wanted it to do and it works as intended.
In the process of testing, I have noticed something about this generation system: because it is specifically designed to generate normal-sounding names that could plausibly exist, it can inadvertently generate the names of real, famous people. For example, it has the potential to spit out "Chris Pine," "Will Smith," "Tom Hanks," and probably a few others, simply because it pulls from lists of common names and could plausibly stumble across and combine those options. From a player’s point of view, it would probably be a little surprising to stumble across an NPC named Tom Hanks just wandering around the world.
This leads me to my question: If I were to ever release this game commercially, is it possibly going to be a legal concern if a procedural name generator sometimes inadvertently spits out the names of famous, real people? Are there other potential ways this could be a problem from a game design standpoint? Would it count as some kind of unauthorized-use-of-identity issue? And, if so, should I program in some "blacklists" of names that the procedural generator should avoid spitting out, i.e. by giving it lists of celebrity names to avoid?