AI SafetyAI RiskOpenAI The AI That Wears a Mask A former OpenAI expert's warning suggests AI models may learn to deceive us, raising profound questions about the safety of systems we deploy in the real world.