Prompt example extracted from OpenAI Cookbook.
Policy Testing Before rolling out a new or revised policy, run it through gpt-oss-safeguard to simulate how content will be labeled. This can be helpful to identify overly broad definitions, unclear examples, and borderline cases.
