https://www.recordedfuture.com/research/artificial-eyes-generative-ai-chinas-military-intelligence
China’s People’s Liberation Army has demonstrated clear interest in leveraging generative artificial intelligence to enhance military intelligence capabilities, according to a new report from Recorded Future’s Insikt Group. The PLA has designed specialized systems that apply generative AI to intelligence tasks and has likely procured AI technology for intelligence purposes, with both military and defense industry entities adapting foreign and domestic large language models to develop specialized tools for intelligence work. These AI-powered systems can reportedly process and analyze intelligence data, generate intelligence products, provide recommendations, facilitate early warning capabilities, and support military decision-making while aiming to improve the speed, efficiency, and scale of intelligence operations.
The PLA’s approach to generative AI integration reveals a sophisticated understanding of both the technology’s potential and its limitations. Patent applications filed by PLA researchers detail methods for using generative AI in open-source intelligence collection, satellite imagery processing, and event data analysis, with one December 2024 patent proposing the use of multiple intelligence disciplines to train specialized military language models. The military has likely adopted models from various sources, including foreign platforms like Meta’s Llama and OpenAI, alongside domestic alternatives from DeepSeek, Tsinghua University, and Alibaba Cloud. Researchers affiliated with the Academy of Military Science have expressed particular optimism about AI’s transformative potential for intelligence research while acknowledging serious challenges including hallucination issues and reliability concerns.
The report highlights significant strategic implications for both China and Western nations as the PLA integrates generative AI into intelligence workflows. Chinese military researchers have recognized counterintelligence risks, warning that foreign adversaries could exploit deepfake technology and AI-generated disinformation to mislead Chinese intelligence personnel, while simultaneously creating similar capabilities that could be used to deceive Western analysts. The extent of successful AI integration remains unclear, as the PLA must navigate challenges including ideological bias in AI training data, the need for specialized military-focused models, and the requirement to maintain human oversight to prevent inaccurate intelligence from degrading decision-making quality in critical military operations.