CYA move. If some bright spark decides to consult Dr. ChatGPT without input from a human M.D., and fucks their shit up as a result, OpenAI can say "not our responsibility, as that's actually against our ToS."