An updated Model Spec for more capable and beneficial AI

OpenAI released an updated Model Spec on Wednesday, outlining five core values to guide more capable AI toward sound judgment, human agency, and faithfulness to human intent. The framework aims to shape AI behavior as models become more powerful, emphasizing alignment with human values and safety.