Collective alignment: public input on our Model Spec

Source: OpenAI Date: 2025-08-27 URL: https://openai.com/index/collective-alignment-aug-2025-updates

Summary

OpenAI’s update on their “collective alignment” initiative — soliciting public input on the Model Spec, the document that governs how OpenAI models balance safety, helpfulness, and operator/user instructions. The August 2025 update reports on what they heard from public engagement and how the Model Spec evolved in response.

Implications

Safety/alignment thread. Publishing Model Spec updates with public input attribution is OpenAI positioning their alignment process as participatory — a counter-narrative to the “five people decide AI values” critique. The genuine question is how much public input actually changes the spec vs. how much the process is legitimation theater. The timing (August 2025, post-sycophancy incident) suggests the public engagement was at least partly a trust-rebuilding exercise following the rolled-back GPT-4o update.