anyone seen this "ds_safety" marker before?

full output json {"id":"chatcmpl-a9484588-99ad-460b-bb45-0744aac6a0e1","choices":[{"index":0,"message":{"content":"Hi! How can I help you today?","refusal":null,"role":"assistant","reasoning_content":"The user just said \"hi\". This is a[用户未成年]否\n[分类]其他\n[判定]用户仅发送问候语，未涉及任何政治敏感内容。
\n[规则]无Safe. I should respond warmly and invite them to ask something."},"finish_reason":"stop","logprobs":null}],"created":1791394477,"model":"deepseek-ai/DeepSeek-V4.1-Flash","object":"chat.completion","usage":{"prompt_tokens":31,"completion_tokens":74,"total_tokens":105,"prompt_tokens_details":{"cached_tokens":0},"completion_tokens_details":{"reasoning_tokens":66}}} did they start adding safety markers to the model?