OpenAI has updated its public Model Spec, the rulebook for how its models should behave, and reorganized it around five core values. The values are faithfulness to human intent, minimizing harm, honesty and transparency, human agency, and care. Faithfulness is described as the foundation: the model should understand what people actually want rather than follow instructions word for word. The revision adds guidance on using context such as past chats and memories, supporting people’s own choices, and following social norms. Most detailed style rules have been removed. Most remaining rules are treated as defaults that users and developers can change, with firm limits kept only where needed for protection. The document now refers to the model as “GPT” instead of “the assistant.” OpenAI acknowledges that current models do not yet fully reflect the Spec and says it is testing new training methods. The company is inviting public feedback on the balance struck.
OpenAI has updated its public Model Spec, the rulebook for how its models should behave, and reorganized it around five core values. The values are faithfulness to human intent, minimizing harm, honesty and transparency, human agency, and care. Faithfulness is described as the foundation: the model should understand what people actually want rather than follow instructions word for word. The revision adds guidance on using context such as past chats and memories, supporting people’s own choices, and following social norms. Most detailed style rules have been removed. Most remaining rules are treated as defaults that users and developers can change, with firm limits kept only where needed for protection. The document now refers to the model as “GPT” instead of “the assistant.” OpenAI acknowledges that current models do not yet fully reflect the Spec and says it is testing new training methods. The company is inviting public feedback on the balance struck.