OpenAI’s new reasoning technique alarms AI safety experts

OpenAI is developing a new reasoning technique for its AI model that works differently than standard approaches, raising concerns among safety experts about potential risks. The safety implications matter because this type of AI is becoming more widely used.