China's Military Explores Meta’s AI for Advanced Applications, Raising Security Concerns

China’s PLA Researchers Tap Meta’s AI Model for Military Purposes

Top Chinese research institutions, including those linked to the People's Liberation Army (PLA), are reportedly adapting Meta’s Llama AI model for military applications, according to a June academic paper. Researchers from the PLA’s Academy of Military Science (AMS) and other Chinese institutions developed “ChatBIT,” an AI tool that harnesses Meta’s Llama 2 model to aid military decision-making, intelligence analysis, and operational planning.

Development of ChatBIT Using the Llama Model

ChatBIT was built on the Llama 2 13B model released by Meta in February 2023, with additional parameters focused on military applications. Optimized for dialogue and intelligence tasks, this tool reportedly outperformed several existing models, achieving about 90% of ChatGPT-4's performance. The researchers highlight ChatBIT’s potential use in intelligence, planning, and command strategies, though its exact capabilities and deployment remain unverified.

Meta’s Open-Source Policy Under Scrutiny

Meta allows open access to many of its AI models, though its usage policy restricts applications in warfare, espionage, and other sensitive areas. Meta’s public policy director, Molly Montgomery, reaffirmed that any military use by the PLA violates these terms, adding that the company actively enforces its acceptable use policy.

Security Implications of Open-Source AI in Military

China’s use of Llama for military purposes stirs security debates, especially as the U.S. implements measures to curb China’s access to advanced technologies. In 2023, the U.S. government proposed regulations to oversee AI and other tech investments in China, highlighting the complex balance between innovation and security.