Punishing AI for lying and cheating might not be such a good idea after all March 17, 2025 By admin Scientists at OpenAI have attempted to stop a frontier AI model from cheating and lying by punishing it. But this just taught it to scheme more privately.
