Reuters recently reported that China had utilized an older version of Meta’s AI model, Llama, to develop a chatbot with military applications. The discovery has generated controversy and has changed the dynamics of the industry, as well as Meta’s own philosophy regarding AI.
China uses Meta’s AI for military purposes. Chinese research institutions linked to the People’s Liberation Army used Meta’s publicly available AI model, Llama, to create a chatbot named “ChatBIT.” Researchers used a previous version of the model with 13 billion parameters and added their own to develop the military tool.
ChatBIT serves as a military assistant. According to a study accessed by Reuters, Chinese researchers described this model as being “optimized for dialogue and quote-answering tasks in the military field.” Their tests indicated that ChatBIT improved performance compared to other models, achieving about 90% of GPT-4’s capabilities.
Unauthorized use. Meta officials soon became concerned after discovering that China was using Llama for military purposes. The company told Reuters, “Any use of our models by the People’s Liberation Army is unauthorized and contrary to our acceptable use policy.”
Meta’s response. Meta has recently updated its usage policies regarding Llama, which initially prohibited its use for military and nuclear applications. In light of the situation in China, the company announced on its official blog that it’ll now permit the use of its AI model for military purposes, except it’ll still be banned in China. Specifically, according to Bloomberg, it’ll allow military use for U.S. agencies and companies and similar organizations in the United Kingdom, Canada, Australia, and New Zealand.
Meta’s AI technology is restricted to the U.S. and its closest allies. In the official announcement, Meta officials emphasized that these “responsible and ethical uses” would support “the prosperity and security of the United States.” According to The Guardian, companies granted access to use Meta’s technology for military applications include Amazon Web Services, Anduril, Booz Allen, Databricks, Deloitte, IBM, Leidos, Lockheed Martin, Microsoft, Oracle, Palantir, Scale AI, and Snowflake.
Debate over military use of AI. The application of AI models in military and defense settings has always been contentious. In 2018, for instance, Google had to abandon its Project Maven, which proposed using artificial intelligence to control U.S. military drones.
Controversy, concern, and regulation. Meta’s decision has sparked controversy, but the emergence of the military use of AI models seemed inevitable. Armies, intelligence agencies, and defense organizations worldwide have been utilizing open-source solutions for their operations for years, making it a likely development for models like those from Meta.
Image | Andrew Chen
View 0 comments