Geo News
Community curated by people like you
LatestAICryptoHealthWorld AffairsUS Politics
White House accuses Moonshot AI of distilling Anthropic's Fable model for Kimi K3
00

White House accuses Moonshot AI of distilling Anthropic's Fable model for Kimi K3

Jul 22, 2026

The White House has accused Chinese startup Moonshot AI of covertly "distilling" Anthropic's Fable model to create its new Kimi K3 AI, escalating US-China tech tensions. An official alleges an industrial-scale IP theft operation using methods to evade detection. The US Treasury warns of sanctions, though some researchers question the claim's technical timeline. Moonshot has not commented.

White House distillation accusations

  • ▪Kratsios alleged Moonshot AI developed a sophisticated internal platform to conduct large-scale distillation of U.S. models
  • ▪The operation allegedly used NVIDIA GB300 servers located in Thailand, which are barred from being sold to China by US export rules
  • ▪Some AI researchers have questioned the accusation, noting the short 15-day window between Anthropic's Fable 5 re-release on July 1 and Kimi K3's launch on July 16
  • ▪The platform allegedly used methods to evade detection, such as rotating access routes
  • ▪Michael Kratsios, Director of the White House Office of Science and Technology Policy, accused Moonshot AI of secretly distilling Anthropic's Fable model to build its Kimi K3 model

Model distillation technique

  • ▪While common and legal when done openly, the White House considers large-scale, covert distillation to be unacceptable theft of U.S. technology
  • ▪Model distillation is a technique where a new AI model is trained on the outputs of a more powerful one

Anthropic tracking findings

  • ▪In a February 2026 report, Anthropic stated it had traced over 3.4 million Claude model exchanges to Moonshot AI
  • ▪Anthropic's report claimed the traffic came from hundreds of fabricated accounts, some of which matched the public profiles of senior Moonshot staff

Kimi K3 model release

  • ▪Anthropic has warned that copied models can lose the safety guardrails of the original and spread uncontrollably once open-sourced
  • ▪Moonshot AI launched its Kimi K3 model on July 16, 2026
  • ▪With 2.8 trillion parameters, Kimi K3 is described as the largest open-weight model released to date
  • ▪The full model weights for Kimi K3 are scheduled to be released to the public by July 27, 2026

US-China AI tensions

  • ▪US Treasury Secretary Scott Bessent warned that covert, industrial-scale distillation attacks could result in sanctions and Entity List designations
  • ▪Moonshot AI has remained silent on the accusations, while Beijing previously called Anthropic's February 2026 findings "groundless."
  • ▪In January 2025, a similar accusation was made against Chinese firm DeepSeek for allegedly copying OpenAI's models

3 sources

Beincrypto
China’s Moonshot Reportedly Stole U.S. AI Tech for Kimi K3
View source article
Cointelegraph
US Accuses Moonshot AI of Covert Anthropic Model Distillation
View source article
Cryptopolitan
White House accuses Moonshot of distilling Anthropic's Fable for Kimi K3 - Cryptopolitan
View source article

Featured stories

View more in China

OpenAI and Anthropic CEOs called to appear at Australian AI inquiry

Sep 27, 2026 · 2 sources

Trump and Xi end Washington summit without major AI agreement

Sep 26, 2026 · 7 sources

Bill Gates warns AI powerful enough to cause a billion deaths

Sep 25, 2026 · 5 sources

FTC opens investigation into OpenAI and Anthropic over consumer protection

Sep 30, 2026 · 7 sources

Story comments

Loading comments…

Related entities

ChinaKimi

Related Projects

Moonshot AIAnthropic

Topics

ChinaAGI arms race & geopoliticsAI RegulationAI copyright & training dataModel distillationAI security

Featured stories

View more in China

OpenAI and Anthropic CEOs called to appear at Australian AI inquiry

Sep 27, 2026 · 2 sources

Trump and Xi end Washington summit without major AI agreement

Sep 26, 2026 · 7 sources

Bill Gates warns AI powerful enough to cause a billion deaths

Sep 25, 2026 · 5 sources

FTC opens investigation into OpenAI and Anthropic over consumer protection

Sep 30, 2026 · 7 sources