China's Moonshot AI Accused of Intellectual Property Theft
· news
China’s Moonshot AI Stole from Anthropic, Trump Tech Adviser Says
The US government has increased scrutiny of Chinese artificial intelligence (AI) companies after allegations that China’s Moonshot AI stole capabilities from top US models. Michael Kratsios, science and technology adviser to former President Donald Trump, accused Moonshot’s AI of using distillation attacks to extract information from Anthropic’s Fable AI.
Distillation attacks are not new; they have been a concern for policymakers since 2022, when the government restricted the export of Nvidia’s most advanced chips over fears that they could be used by the Chinese military. Governments worldwide have since cracked down on the smuggling of high-tech components.
The case of Anthropic’s Claude AI models, allegedly illicitly extracted by Alibaba in June using distillation attacks, highlights the willingness of foreign companies to push intellectual property boundaries. In an era where open-source AI is increasingly popular, it is clear that not all players adhere to the same rules.
Moonshot’s Kimi K3 model, unveiled last week and touted as rivaling top US technology, will be released as an open-source model on July 27th. This raises questions about its true origins and whether it has been built on stolen tech. Treasury Secretary Scott Bessent has vowed to examine whether Chinese AI models have stolen capabilities from American rivals and hinted at possible sanctions.
The increased focus on AI intellectual property theft comes amid rising tensions between Washington and Beijing ahead of President Trump’s meeting with Chinese counterpart Xi Jinping in September. The stakes are high, and it remains to be seen how this latest development will impact the diplomatic landscape.
As the world waits for China’s response to these allegations, one thing is clear: the lines between innovation and theft have never been more blurred. With open-source AI models becoming increasingly popular, companies can mask their true intentions with greater ease. The US government is committed to protecting American intellectual property at all costs, but only time will tell if China’s rapid advancements in AI are built on a foundation of stolen tech or genuine innovation.
Reader Views
- CMColumnist M. Reid · opinion columnist
While the focus on Moonshot's alleged theft of intellectual property from Anthropic's AI models is understandable, it's worth noting that this case highlights a larger issue: the lack of international standards for AI development and usage. As AI becomes increasingly integrated into daily life, we risk creating a world where technological advancements are viewed as a zero-sum game, where one country's gain must come at another's expense. To truly address intellectual property theft in the AI sector, policymakers need to work towards establishing clear guidelines and cooperation with international partners, rather than simply placing blame.
- ADAnalyst D. Park · policy analyst
The Moonshot AI debacle highlights a critical concern: China's reliance on intellectual property theft as a business model rather than innovation. While distillation attacks have been a known threat for years, policymakers have largely overlooked their application in AI development. The real question is not just whether Moonshot's Kimi K3 was built on stolen tech, but how deep-rooted this practice is within the Chinese AI ecosystem. Without meaningful reforms, the US risks perpetuating a cycle of dependence on intellectual property protections rather than fostering genuine competition and cooperation in AI research.
- RJReporter J. Avery · staff reporter
The AI industry's Wild West is getting some much-needed scrutiny. While it's understandable that companies like Anthropic and Moonshot are pushing the boundaries of innovation, the use of distillation attacks to extract proprietary information from rivals raises serious red flags about IP theft. The real question is what this means for the long-term sustainability of open-source AI models, which rely on collaboration and trust among developers. If foreign entities are willing to skirt ethics and steal tech, does it undermine the entire concept?