Objective To analyze Weibo "Tree Hole" users′ suicidal ideation related information, and to provide reference for suicide early warning and intervention using social media. Methods The "Zou Fan" Weibo tree hole was 24 hours monitored by an artificial intelligence robot from August to October in 2018, the leaving massage on suicidal ideation was screened according to knowledge mapping, and the level 6-10 high risk suicidal ideation information was automatically distinguished, then the suicide risk was categorized. Results Totally 118,000 messages from "Zou Fan" Weibo tree hole were obtained by the robot. The peak time period for writing these messages was 22:00 to 2:00 in the next morning, and 711 messages were identified as level 6-10 high risk suicide, and its users were mainly focused in 16-26 years old, and their expressed primary suicide methods were jumping off the building, cutting wrists, charcoal-burning, hanging and jumping into a river. Conclusion It is a more and more popular phenomenon to express suicidal ideation via social media, the artificial intelligence technology could be used to identify and rescue high risk suicide population. Suicide prevention should improve the 24 h crisis warning and intervention mechanism, pay attention to adolescents′mental health, and create a good network social environment.