A senior White House adviser has accused Beijing-based AI startup Moonshot AI of “large-scale” intellectual property theft, alleging the company used a technique called distillation to extract capabilities from Anthropic’s Fable AI model to develop its Kimi K3 system. President Donald Trump, stated in a social media post that Moonshot AI “distilled” Anthropic’s Fable to build its K3 model, according to sources 1, 4, and 5. Kratsios also claimed Moonshot gained access to restricted Nvidia GB300 servers—high-performance AI chips banned from sale to China—to train its models.
White House Accuses China’s Moonshot AI of Stealing Anthropic Tech for Kimi K3
The allegations come amid heightened U.S. scrutiny of Chinese AI firms, with Treasury Secretary Scott Bessent previously warning that the U.S. would sanction foreign AI models found to have stolen American technology. The White House has not yet announced specific penalties against Moonshot AI, and the company has not publicly responded to the claims, as noted in source 4.
NIST Report Highlights U.S. AI Superiority, but Chinese Models Gain Ground
A recent National Institute of Science and Technology (NIST) study compared U.S. and Chinese AI models, finding that OpenAI and Anthropic outperformed DeepSeek across 19 benchmarks. U.S. Commerce Secretary Howard Lutnick highlighted the results, stating that DeepSeek lags far behind
in areas like cyber and software engineering. The tests, conducted under the newly established Center for AI Standards and Innovation (CAISI), included evaluations of models such as OpenAI’s GPT-5, GPT-5-mini, and GPT-oss, as well as Anthropic’s Opus 4.
Despite this, Chinese AI models like Moonshot’s Kimi K3 are challenging U.S. dominance. Kimi K3, a 2.8-trillion-parameter open-weight model, was unveiled in July 2026 and is described as the largest such model released to date. While it trails Anthropic’s Claude Fable 5 and OpenAI’s GPT 5.6 Sol in some rankings, it has sparked debates over the global AI leadership race, according to source 3. The model’s full code is set to be released on July 27, a date that could test its capabilities against U.S. systems.
Alleged Distillation Campaign and Chip Access Raise Security Concerns
In a February report, Anthropic identified over 16 million Claude exchanges linked to Moonshot, with traffic originating from hundreds of fake accounts tied to senior Moonshot staff. The firm also alleged that Moonshot targeted Fable’s reasoning, coding, and vision skills, as detailed in source 4.
Kratsios added that Moonshot allegedly used Nvidia GB300 servers in Thailand—chips subject to U.S. export controls. Washington has tightened restrictions on AI chip sales to China since May 2026, with a June raid on a Super Micro office in Taiwan targeting alleged chip smuggling. These actions underscore growing U.S. efforts to curb Chinese access to advanced AI infrastructure, as reported in source 4.
Kimi K3’s Market Impact and Broader Implications
The release of Kimi K3 sent shockwaves through global markets, with Asian AI competitors like Z.ai and MiniMax Group experiencing steep stock declines. Alibaba, a Moonshot backer, also fell 4%, while the Nasdaq 100 futures dipped. Analysts noted that Kimi K3’s pricing strategy—near Anthropic’s mid-range tier—positions it as a cost-effective alternative to U.S. models. However, benchmarks show it still lags behind leading U.S. systems, according to source 3.
The U.S. maintains a significant spending advantage in AI, with $190 billion in investments projected for 2026. Yet Chinese models continue to close the gap, raising concerns about technology transfer and national security.
U.S.-China AI Rivalry Intensifies Ahead of Diplomatic Talks
The accusations against Moonshot AI coincide with heightened diplomatic tensions between the U.S. and China. Trump is expected to meet Chinese President Xi Jinping in September, with AI policy likely to be a key topic. Meanwhile, Chinese officials emphasize open access to AI, with Xi Jinping stating that the technology should not be a solo performance by a single country
during a recent AI summit. In contrast, the U.S. has expanded public access to models like OpenAI’s GPT-5.6 Sol, reflecting differing approaches to AI governance.
As the U.S. weighs potential sanctions and regulatory actions, the case of Moonshot AI underscores the broader struggle over AI leadership. With Kimi K3’s full release approaching, the coming weeks could define the trajectory of the global AI race.
bbc.com | tomshardware.com | cryptopolitan.com | beincrypto.com | seekingalpha.com
Related reading
