EVERY SIGNAL, SORTED.

United States, OpenAI, Model Spec, Model Behavior, Human Intent, Harm Reduction, Transparency, Human Agency, Memory, Oversight, Side Effects, Product Evaluation and Regression Testing

OpenAI rewrote the Model Spec around five values. Show where the product still disagrees

OpenAI reorganized its public behavior specification around human intent, harm, honesty, agency and care. The useful next artifact is a product-level map showing which scenarios match the document, which do not and which version is actually running.