In order to further improve the legal regulation effect of generative artificial intelligence, this paper constructs a bioterrorism threat information fingerprint matching module based on hypergraph network. This module uses two key sub-modules (relationship matching learning module and multi-scale information construction module) to simultaneously analyze pairwise relationships and triple-based relationships, so as to achieve text semantic matching of generative artificial intelligence, improve its risk judgment effect, and further improve the screening of artificial intelligence legal risks. Moreover, this paper combines experimental analysis to analyze the semantic matching of generative artificial intelligence texts, verifies the reliability of the method model, and discusses the legal risks of artificial intelligence and its regulatory path after the experiment, which provides a more reliable technical support for the subsequent generative artificial intelligence legal risks.