Judgment on sensitive content cases
Practice working through borderline examples, such as a fiction request that could also be a real-world capability question, and write out the reasoning behind your decision.
The AI Model Policy Trainer role centers on close cases where content is ambiguous between legitimate and harmful uses.
Source posting