Help researchers develop more secure AI.
Or are we helping researchers develop more sophisticated AI attacks?
Edit: this is one of those things we should have regulations for.. to have someone ask, "why exactly are you doing this?" and act appropriately