Key facts
- US officials accuse Chinese AI firm Moonshot of copying Anthropic's Fable LLM.
- The accusation involves the use of restricted chips.
- Experts question whether model distillation can account for the rapid progress of Moonshot's Kimi K3.
- The claims suggest a potential technological theft and raise questions about AI development practices.
US officials have accused Chinese AI firm Moonshot of illicitly copying Anthropic's Fable large language model (LLM) by utilizing restricted chips. The allegations suggest a potential case of technological theft, with officials pointing to the rapid advancement of Moonshot's Kimi K3 model as evidence.
However, the claims have been met with skepticism from global AI experts. These experts question the feasibility of achieving such rapid progress solely through model distillation, a process where a smaller model learns from a larger one. The experts' pushback suggests that the US officials' assertions may not fully account for the technical capabilities demonstrated by Moonshot's Kimi K3.