I believe it depends on the exploit and intent. Let me make an example.
In World of Warcraft, there might be a boss in the latest raid dungeon. Raid Group A notices that the boss is weak to a certain tactic (eg. kiting) and uses this to help mitigate some of the damage recieved. As this is a new boss which no one has fought before, it is easily argued that kiting this boss could be considered a legitimate tactic. It is, however, an oversight in Blizzard's code and thus could also be deemed an exploit. As Raid Group A is the first to ever fight this boss outside of Blizzard's closed testing facilities, how are they to know?
In that respect, they might be exploiting the boss - the nature of it (that being that it is unknown) and the intent (honestly believing that this could be 'the' tactic to beat the boss) however I would find myself hard pressed to say that they are in the wrong and should be punished for it.
Example 2:
Raid Group B is facing up against a gauntlet-type encounter where they need to rush to the end and hit a trigger to continue/finish the encounter. A member of Raid Group B notices that if they jump up on the terrain in a certain way, they can bypass the gauntlet entirely without recieving a single bit of damage. Raid Group B notices this, and uses it to get past the gauntlet encounter and recieve their phat lewtz.
Exploiting the terrain in such a way, even if it was previously undiscovered, is unquestionably a bug and an unintended use of an oversight by the programmers, and thus could be deemed an exploit which - even though previously unknown - still warranted punishment as the other nature (jumping over terrain to skip an encounter) is clearly an Exploit.
It's not as clear cut as "Exploit=Ban". One needs to look at the deeper mechanics, intent and knowledge behind the exploit before dishing out punishment or you will catch a lot of innocents in your wide net.
