White House Accuses Chinese AI Startup of Stealing Anthropic Technology

The White House technology advisor, Michael Kratsios, accused Chinese startup Moonshot AI of stealing technology from Anthropic, a US AI company, to develop its Kimi K3 model. Kratsios claimed that Moonshot AI used a process called 'distillation' on Anthropic's Fable 5 model, which involves training one system using data from another. He alleged that Moonshot AI conducted a large-scale operation to rapidly switch model access to avoid detection. Anthropic's Fable 5, its most powerful model, was launched in early June and temporarily suspended at the US government's request before reopening in early July. Kimi K3, released on Thursday, is an open-source model that is free to download and adapt, and it has performed well in benchmarks, particularly in application programming. Kratsios stated that the Trump administration will investigate and may impose sanctions if foreign models are found to be stealing US technology. Treasury Secretary Scott Bessent echoed these concerns, saying traces of US models have been found in several Chinese AI systems.

Global Impact

This accusation deepens the US-China technology rivalry, with potential economic and geopolitical consequences. Economically, it could lead to sanctions or trade restrictions that disrupt the global AI supply chain, affecting companies like Nvidia and other chipmakers.