Chinese military researchers had analyzed U.S. efforts to apply generative AI to military intelligence task, according to a new report.
China’s People’s Liberation Army (PLA) is actively pursuing generative artificial intelligence (AI) capabilities to gather and analyze military intelligence, according to a threat analysis report published on June 17.
U.S.-based Insikt Group, the research division of threat intelligence consultancy Recorded Future, examined articles in PLA-affiliated media, research papers, patent applications, and procurement orders, and said the material showed the Chinese military’s “clear interest” in generative AI tools.
The report said the Chinese regime’s military-industrial complex is likely using large language models (LLMs) from a number of foreign companies, including Meta, OpenAI, Google DeepMind, Anthropic, BigScience, and domestic entities such as DeepSeek, Tsinghua University, Zhipu AI, and Alibaba Cloud.
The U.S. military is also investing in AI tools to improve its decision-making process and other warfighting capabilities. According to a Defense Department contracts document published on June 16, the Pentagon awarded OpenAI a $200 million contract to develop “prototype frontier AI capabilities to address critical national security challenges in both warfighting and enterprise domains.”
Throughout last year, U.S. Marines deployed to the Pacific tested a generative AI tool developed by Vannevar Labs, using the tool to sift through foreign intelligence.
Authors of the Insikt Group report said PLA researchers had analyzed U.S. efforts to apply generative AI to military intelligence tasks, likely aiming to learn from the experience.
The report said that PLA media and researchers have argued for the use of generative AI, citing its potential benefits, including the enhancement of intelligence gathering and analysis, while recognizing challenges and risks, such as inaccurate intelligence.
If PLA intelligence analysts use generative AI models aligned with Chinese Communist Party (CCP) ideology or are “trained on ideologically biased analytical products, the PLA risks reducing the objectivity of intelligence analysis,” it said.
For the West, the report said, the PLA’s adoption of generative AI tools could create “technology transfer challenges” and highlight the risk of AI-generated disinformation.
By Lily Zhou
