Not when powerful AI is being fine-tuned to maximize a reward (money).

This is the whole reinforcement learning alignment problem, just with human+AI instead of AI by itself. Unaligned incentives (money vs. human well-being).