US officials allege Moonshot AI copied Anthropic model as Kimi K3 weights released

US officials allege Moonshot AI copied Anthropic model as Kimi K3 weights released

TechCrunch reported that U. S. officials alleged Beijing-based Moonshot AI copied Anthropic’s Fable large-language model using restricted Nvidia chips, a claim experts said was technically unlikely. Earlier in July, Bloomberg noted that Moonshot AI made the 2. 8-trillion-parameter Kimi K3 model weights available for free download, while VentureBeat added that the accompanying license imposed extra obligations on large enterprises and model-as-a-service providers. ABC News said the company temporarily halted new subscriptions for Kimi K3 after demand quickly outstripped its computing capacity. Tech Xplore highlighted that the model topped Arena’s coding benchmark, drawing comparisons with Anthropic’s Claude and OpenAI’s ChatGPT. TechCrunch earlier reported that U.

S. officials were weighing possible restrictions on open-weight Chinese models over security and competition concerns. PBS said the White House was assessing the low-cost Chinese models as a competitive threat, and CNBC highlighted a debate among U. S. officials and AI firms over how to regulate model distillation. AP observed that U. S. companies and developers were increasingly adopting Chinese AI models because of comparable performance at lower cost.

参与者

Moonshot Kimi K3 Anastasios Angelopoulos Xi Jinping World Artificial Intelligence Conference Huawei Bank of America Anthropic OpenAI Yang Zhilin Russ Salakhutdinov

地点

Beijing, China Shanghai, China

文章

2026年七月28日
1 篇总计
Kimi K3's full weights are here, but they're 'open' with a caveat: What enterprises should know
Kimi K3's full weights are here, but they're 'open' with a caveat: What enterprises should know

VentureBeat • 07月28日 02:32

Moonshot AI has made the 2.8-trillion-parameter Kimi K3 model weights publicly available, but its license imposes extra obligations on large enterprises and model-as-a-service providers.

可信度 78% 操纵 9% 中立
2026年七月27日
1 篇总计
China’s Moonshot Releases Breakthrough AI Model for Download
China’s Moonshot Releases Breakthrough AI Model for Download

Bloomberg • 07月27日 15:33

Beijing-based Moonshot AI has made the weights of its Kimi K3 large language model available for free download, allowing developers to modify and host the system themselves.

可信度 78% 操纵 12% 中立
2026年七月26日
1 篇总计
Cheaper, intelligent Chinese AI models make inroads in the US
Cheaper, intelligent Chinese AI models make inroads in the US

AP News • 07月26日 04:05

U.S. companies and developers are increasingly adopting AI models from Chinese firms such as Moonshot, DeepSeek and Z.ai because they offer comparable capabilities to leading U.S. systems at a fraction of the price.

可信度 78% 操纵 10% 中立
2026年七月25日
1 篇总计
From Silicon Valley to DC, tech world obsessed with AI distillation
From Silicon Valley to DC, tech world obsessed with AI distillation

CNBC • 07月25日 12:00

U.S. officials and leading AI companies are weighing how to address AI distillation, a technique that allows smaller models to learn from larger ones, after Chinese labs released competitive open-weight models.

可信度 73% 操纵 9% 中立
2026年七月23日
2 篇总计
Experts say exploiting Anthropic's Fable isn't how Kimi K3 got so good
Experts say exploiting Anthropic's Fable isn't how Kimi K3 got so good

TechCrunch • 07月23日 11:00

White House science adviser Michael Kratsios claimed Moonshot replicated Anthropic’s Fable LLM using prohibited chips, but AI researchers say the timeline and technical hurdles make such distillation unlikely.

可信度 68% 操纵 35% 中立
China's Moonshot AI stole from Anthropic, Trump tech adviser says
China's Moonshot AI stole from Anthropic, Trump tech adviser says

BBC • 07月23日 00:52

A White House technology adviser claimed that China's Moonshot AI extracted capabilities from Anthropic's Fable model and used restricted Nvidia hardware, prompting U.S. officials to consider sanctions.

可信度 71% 操纵 12% 中立
2026年七月22日
1 篇总计
Breakthroughs in Chinese AI models threaten U.S. lead in tech race
Breakthroughs in Chinese AI models threaten U.S. lead in tech race

PBS • 07月22日 22:35

Open-weight AI models from Chinese firms Moonshot AI and Alibaba are reported to approach the performance of leading U.S. systems while costing far less, prompting the White House to consider policy actions and a shift in federal AI research funding.

可信度 72% 操纵 46% 中立
2026年七月21日
2 篇总计
America needs to stop getting shocked by Chinese AI
America needs to stop getting shocked by Chinese AI

The Verge • 07月21日 11:08

Beijing-based Moonshot AI and Alibaba introduced new large language models that claim performance near top US systems and are priced lower, prompting market and policy reactions in the United States.

可信度 78% 操纵 14% 中立
China's new AI model halts new subscriptions as demand swamps capacity
China's new AI model halts new subscriptions as demand swamps capacity

ABC News - Breaking News, Latest News and Videos • 07月21日 09:00

Moonshot AI has temporarily halted new subscriptions for its Kimi K3 model after demand quickly exceeded the platform's computing capacity.

可信度 68% 操纵 12% 中立
2026年七月20日
2 篇总计
OpenAI is scared of open-weight models. Should the US be?
OpenAI is scared of open-weight models. Should the US be?

TechCrunch • 07月20日 19:33

U.S. officials are weighing restrictions on Chinese open-weight large language models amid concerns about competition, security and market impact, while industry leaders argue such limits could hinder innovation.

可信度 68% 操纵 32% 中立
Americans Are Freaking Out Over China's Open-Source AI Strategy
Americans Are Freaking Out Over China's Open-Source AI Strategy

Business Insider • 07月20日 04:41

China's Moonshot AI released the Kimi K3 model, prompting US tech leaders to discuss the risks and competitive implications of open-weight artificial intelligence.

可信度 72% 操纵 32% 中立
2026年七月18日
1 篇总计
Chinese AI model takes US tech industry by surprise with abilities rivaling Claude and ChatGPT
Chinese AI model takes US tech industry by surprise with abilities rivaling Claude and ChatGPT

Tech Xplore • 07月18日 08:06

Moonshot's Kimi K3 large-language model performed at the top of Arena's coding benchmark, prompting comparisons with Anthropic's Claude and OpenAI's ChatGPT, while Chinese officials highlighted the need for global AI cooperation.

可信度 72% 操纵 12% 中立