Meta’s Llama AI Potentially Misused by China’s Military, Attaining 90% of ChatGPT-4’s Power
AI-generated Image
Image Credit: Jacky Lee
Recent revelations indicate that the People’s Liberation Army of China has been leveraging open-source artificial intelligence models, including Meta’s Llama, alongside domestic counterparts like Baidu’s Wenxin Yiyan and Tsinghua University’s ChatGLM, to bolster its military research and operational capabilities.
[Read More: Meta Unleashes Llama 3.1: The Most Powerful AI - And It’s Free?]
The People's Liberation Army
The People's Liberation Army (PLA), established on August 1, 1927, serves as the unified military force of the People's Republic of China and the armed wing of the Chinese Communist Party (CCP). Comprising the Ground Force, Navy, Air Force, and Rocket Force, the PLA operates under the direct command of the Central Military Commission, chaired by CCP General Secretary Xi Jinping. In recent decades, the PLA has undergone significant modernization, focusing on technological advancements and enhanced combat readiness, including the development of advanced missile systems, fifth-generation fighter jets, and the expansion of naval capabilities. Its primary mission is to safeguard China's sovereignty, territorial integrity, and national interests, projecting power in the Asia-Pacific region, particularly concerning Taiwan and territorial disputes in the South China Sea.
[Read More: Can China Bypass OpenAI's Restriction?]
Integration of Global and Domestic AI Technologies
Extensive examination of Chinese official research documents has confirmed that the PLA is not only utilizing Meta’s Llama but also integrating domestic AI models such as Wenxin Yiyan and ChatGLM. By combining these with foreign AI technologies, the PLA is developing specialized “military operational models”. These models are designed for real-time battlefield deployment, mechanical control, data organization, and enemy assessment, aiming to enhance combat simulations and strategize against U.S. military deployments.
[Read More: Superintelligence: Is Humanity's Future Shaped by AI Risks and Ambitions?]
Meta’s Llama and PLA’s Military Advancements
Meta's Llama models are subject to an Acceptable Use Policy that explicitly prohibits their use in military applications. The policy states that users must not engage in activities related to "military, warfare, nuclear industries or applications, espionage," among other restricted areas. However, a recent study by The Jamestown Foundation highlights concerns that China may have improperly accessed Meta’s Llama model. The study estimates that the capabilities of PLA’s adapted Llama model reach approximately 90% of OpenAI’s ChatGPT-4. This adaptation is believed to aid the PLA in intelligent analysis, strategic planning, training, and command decisions. Additionally, the technology is expected to assist Chinese police forces in maintaining stability and enhancing surveillance operations.
The Jamestown Foundation
The Jamestown Foundation, established in 1984, is a Washington, D.C.–based non-profit research institute dedicated to informing policymakers and the public about events and trends in regions critical to the United States and its allies. Initially focused on supporting Soviet defectors, the foundation now concentrates on areas such as China, Eurasia, Russia, and global terrorism. It publishes several periodicals, including China Brief, Eurasia Daily Monitor, and Terrorism Monitor, providing in-depth analyses of geopolitical developments. The foundation is recognized for its objective reporting and high factual accuracy, earning a "Least Biased" rating from Media Bias/Fact Check.
Strategic Focus on AI-Driven Warfare
Chinese President Xi Jinping has been a strong proponent of enhancing military strength through intelligent technologies. Over the past five years, Xi has emphasized the importance of “intelligent warfare”, making the development of AI-driven military models a key national strategy. The 2020 PLA joint operations plan outlined ambitions to use AI for better coordination and faster military responses, highlighting the preference for machine-operated forces over human-led ones for quicker detection and reaction to battlefield changes.
[Read More: AI Arms Race: Navigating the Tumultuous Waters of Global Tech Dominance]
Development of Proprietary Military AI Models
Chinese military and research institutions have been actively developing proprietary AI models tailored for defense applications. Companies like Yuanting Technology have secured substantial funding to advance military models such as Tianji, which address the specific needs and challenges of defense-related AI applications. These models incorporate multiple open-source bases, including Llama and ChatGLM, and are fine-tuned to create domain-specific solutions for military use.
[Read More: AI on the Battlefield: A New Era of Warfare]
Academic Contributions and Technological Innovations
Institutions like the People’s Armed Police Engineering University have published research on military AI applications, demonstrating how large language models can be utilized for battlefield deployment and strategic simulations. These studies showcase the PLA’s efforts to match and potentially surpass U.S. military AI advancements by developing systems that can predict enemy movements and optimize combat strategies based on natural language descriptions of various scenarios.
[Read More: Semiconductor Showdown: How Taiwan’s Chips Shape the AI Race and Geopolitics]
Concerns Over AI in Authoritarian Regimes
Experts warn that the deployment of advanced AI models by authoritarian regimes like China poses significant risks. The ability to use AI for information warfare, surveillance, and maintaining internal control raises ethical and security concerns. The PLA’s integration of AI into its operations underscores the potential for these technologies to be used in ways that could suppress freedoms and enhance state control, highlighting the urgent need for international dialogue and regulation on military AI applications.
[Read More: Hong Kong's Controversial Ban on Protest Anthem: A Blow to Free Speech?]
Source: Radio Free Asia, GitHub, Reuters
