Lead

The Trump administration has accused Chinese artificial intelligence startup Moonshot AI of covertly extracting capabilities from leading US AI models, sparking pushback from global AI experts who say the allegations lack justification. The accusations, leveled by White House science and technology adviser Michael Kratsios and later echoed by Under Secretary of State Jacob Helberg, target Moonshot’s recently released Kimi K3 model, which has drawn attention for its performance and scale.

Separately, Moonshot AI this week launched a “Kimi Ambassador Program” aimed at recruiting entrepreneurs, developers, content creators and university students to build communities and promote the future of AI, as reported by the South China Morning Post.

Coverage Comparison

The South China Morning Post reported on multiple aspects of the story, covering the Trump administration’s accusations, the global AI community’s response, and Moonshot’s ambassador program launch. The outlet’s reporting on the accusations quotes US officials and includes a rebuttal from a Moonshot AI employee, while its coverage of the ambassador program relies on information from Moonshot’s website and direct quotes from the company.

Key Claims

  • The Trump administration accused Moonshot AI of covertly extracting capabilities from leading US artificial intelligence models, specifically targeting Anthropic’s Claude Fable 5 through large-scale “distillation,” as reported by the South China Morning Post. White House science and technology adviser Michael Kratsios made the allegation on X on Wednesday, describing the practice as “stealing proprietary US technology.” Under Secretary of State Jacob Helberg characterized the alleged distillation efforts as “a heist of invaluable American Intellectual Property.”
  • Kratsios did not present evidence that outputs from Fable 5 had been incorporated into Kimi K3, according to the South China Morning Post. The timing of the two models’ releases—Fable 5 went public on July 1 and Kimi K3 launched on July 15—prompted disagreement among AI researchers about whether Moonshot could have used the US model to train its latest system.
  • Global AI experts pushed back against the Trump administration’s claims on X on Thursday, calling the accusations “political” and “reckless,” pointing out that an AI model’s output was not copyrighted, as reported by the South China Morning Post.
  • Moonshot AI employee Randy Xian responded on X, noting the 15-day window between the releases and writing: “Yes, Fable went public on July 1 and K3 launched on July 15. We trained a brand new frontier model in JUST 15 DAYS. Guinness World Record stuff.”
  • Moonshot AI launched the Kimi Ambassador Program on Tuesday, aiming to recruit entrepreneurs, developers, content creators, and university students who are influential in their communities and have integrated Kimi into products or workflows, as reported by the South China Morning Post. Successful candidates will gain early access to unreleased Kimi models and experimental features, API credits, direct engagement with Moonshot’s product team, and opportunities to showcase their work through official channels.
  • Kimi K3 has 2.8 trillion parameters, making it one of the largest AI systems released by a Chinese developer, according to the South China Morning Post.

Perspectives

US Officials: Michael Kratsios, White House science and technology adviser, accused Moonshot AI of conducting “large-scale, covert industrial distillation” against US models, describing the practice as “stealing proprietary US technology.” Under Secretary of State Jacob Helberg echoed the remarks, characterizing the alleged distillation as “a heist of invaluable American Intellectual Property.” The accusations reflect Washington’s concern that Chinese developers may be circumventing US efforts to preserve its AI advantage through restrictions on access to proprietary American models and controls on exports of advanced semiconductors.

Moonshot AI and Global AI Experts: A Moonshot AI employee, Randy Xian, pushed back on the claims by highlighting the narrow 15-day window between the release of Fable 5 and Kimi K3, suggesting the timeline makes large-scale distillation implausible. Global AI experts cited by the South China Morning Post called the accusations “political” and “reckless,” arguing that the practice of using a model’s outputs does not constitute intellectual property theft and that AI model outputs are not copyrighted.