The principle "Microsoft AI Principles" has mentioned the topic "safety" in the following places: