Approaches To Safety In Inverse Reinforcement Learning