OpenAI uses automated systems to enforce its usage policies across all its products, detecting and blocking requests that violate its guidelines before or after model output is produced.

OpenAI applies its usage policies and model specification across ChatGPT, the API, DALL-E, Sora, and enterprise products. The company uses both model-level alignment (training models to refuse harmful requests) and post-generation classifiers to screen outputs. Operators who deploy OpenAI's API must adhere to these policies, and automated systems detect violations.