Debate Over Chinese AI Models Splits Trump Administration

Key points

• The Trump administration is divided on how to address China's AI growth as of October 2023.
• China's Moonshot AI lab has developed the Kimi K3 model, rivaling top US models.
• Anthropic alleged a significant distillation attack by Alibaba on their AI model.

The Trump administration is currently facing an internal debate over how to respond to the fast progress of China's AI models. This disagreement, first reported by Wired AI, involves two main camps: the White House, which is pushing for stricter controls on Chinese AI, and the Commerce Department, which questions the feasibility of such measures.

This situation follows the release of the Kimi K3 model by China's Moonshot AI lab. This new model is reportedly competitive with leading US models from companies like Anthropic and OpenAI. The core of the White House's concern is the practice known as distillation, where Chinese labs may enhance their models by training them on US models. This was brought into focus after Anthropic accused Alibaba of executing a large-scale distillation attack on its Claude Fable 5 model.

The White House is considering a presidential action, though not necessarily an executive order, to address these practices. Treasury Secretary Scott Bessent referred to these distillation tactics as intellectual property theft and hinted at possible sanctions against such covert actions.

While these discussions are ongoing, no formal requests have been made to the Commerce Department for input. The Commerce Department, which controls export regulations, has become a significant player in this debate, partly due to its role in setting industry standards and controls.

To counterbalance China's influence, there have been talks of incentivizing US labs to create their own open weight models. These models are open for use and modification, which could help reduce the risk of hacking attempts on government systems.

What should concerned citizens know?

For ordinary people, the rise of powerful AI models in China means there could be increased competition in the technology sector. However, there is also a potential risk if these models are used for malicious purposes. Keeping an eye on the regulatory measures and understanding the basics of AI model security can help individuals stay informed about how these developments might affect them.

The debate over AI measures highlights the growing concerns about the security and competitive balance between major world powers in the field of artificial intelligence.