The principle "Universal Guidelines for AI" has mentioned the topic "safety" in the following places: