文本内容安全识别不够严?
我测试了一个很严重的违规词汇,竟然都没risky,而是review。 返回值如下,根据id你们应该查得到内容 {"errcode":0,"errmsg":"ok","detail":[{"strategy":"content_model","errcode":0,"suggest":"review","label":20001,"prob":90},{"strategy":"keyword","errcode":0}],"trace_id":"64107881-65f15508-586e63ec","result":{"suggest":"review","label":20001}}