In the face of the challenges posed by the networking and concealment of terrorist activities, this paper systematically explores the development and utilization of open-source intelligence related to terrorism driven by artificial intelligence technology. The research shows that the multimodal data collection technology based on natural language processing and image recognition can achieve precise capture of public sources such as social media and dark web forums. Among them, the AI-driven web crawler has a higher data acquisition efficiency compared to traditional methods. It also proposes to realize distributed intelligence verification through a crowdsourcing model, involving social forces. This provides a theoretical framework and technical implementation route for the paradigm transformation of counter-terrorism intelligence work in the AI era, and has certain practical value for enhancing the country's ability to prevent and control national security risks.