By incorporating social representation theory with science communication and by using a critical milestone scientific crisis (i.e., the scandal of Chinese gene-edited human babies in 2018) as a dividing point, this study adopted a network agenda-setting approach to explore how various actors (i.e., scientists, the media, laypeople, and the government) engaged in the construction of social representations of the controversial issue of gene editing on Chinese social media (i.e., Sina Weibo). Based on large-scale social media data, supervised machine learning was employed to identify attribute categories, and semantic network analysis was used to construct attribute networks. Results reveal that after the 2018 crisis, gene editing received increasing social attention on Chinese social media. Further, two trends emerged in social representations of gene editing on social media: de-scientization and medialization. The following dynamic agenda interactions among various actors were found: On the one hand, the media and laypeople’s attribute network agendas converged while scientists and the media’s diverged after the scandal. This indicates a scientific crisis can serve as a trigger for agenda convergence and divergence among different actors online. On the other hand, there were constant agenda interactions, such as between the Chinese government and the media. This reveals a feature of Chinese science communication—the media not only mediates between scientists and the public, it also observes the government’s agenda closely when representing controversial scientific issues such as gene editing.