Trump adviser accuses Moonshot AI of stealing from Anthropic
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Michael Kratsios, President Trump's Science and Technology adviser, accused Moonshot AI of conducting a "large scale" effort to steal capabilities from top US AI models, in a social media post on Wednesday.
- Kratsios said Moonshot ran the alleged campaign through distillation, where a weaker AI model extracts answers from a stronger one, specifically targeting capabilities from Anthropic for its Kimi K3 model.
- Moonshot AI also reportedly gained access to restricted, cutting-edge Nvidia servers to train its models, according to Kratsios, with the BBC reaching out to Moonshot, Anthropic, the White House, and Nvidia for comment.
- Moonshot's Kimi K3, unveiled last week, drew global attention for appearing to narrow the gap between Western and Chinese AI models; Moonshot itself claims it can rival top US technology.
- Treasury Secretary Scott Bessent, speaking to Fox Business on Tuesday, said the US would examine whether Chinese AI models stole capabilities from American rivals and warned the US has the ability to sanction them if overseas models are found stealing from US companies.
- The US-China diplomatic calendar provides timing context: Trump's increased scrutiny of Chinese AI comes as he is expected to meet Chinese counterpart Xi Jinping in September.
Why it matters: Public accusations from two senior US officials within 24 hours — naming distillation, restricted Nvidia chips, and the threat of sanctions — signal a hardening US position on Chinese AI development right before a Trump-Xi meeting. For Anthropic, the specific naming of its technology in a White House adviser's post puts it at the center of a US-China IP fight; for Nvidia, it raises fresh questions about enforcement of export-controlled chip access.
