In a significant escalation of its approach to foreign artificial intelligence (AI) threats, the US government is reportedly intensifying efforts to counter the activities of foreign entities, particularly those based in China, engaged in what it terms “industrial-scale campaigns” to replicate American AI systems. This development follows a memo from the White House that outlines these concerns, as revealed by the Financial Times.
The document states, “The US government has information indicating that foreign entities, principally based in China, are engaged in deliberate, industrial-scale campaigns to distil US frontier AI systems.” Distillation, in AI terminology, involves training a model using outputs from a more advanced system, enabling it to emulate the original’s capabilities.
Concerns regarding these practices have been underscored by previous accusations from the AI company Anthropic, which claimed that firms such as DeepSeek, Moonshot, and MiniMax were conducting “industrial-scale distillation attacks” on their models. Anthropic’s recent discussions with the White House suggest that the company may have played a role in bringing the distillation issue to the government’s attention.
The White House’s response includes intentions to share relevant information with American AI companies and facilitate coordination against these perceived threats. This move signals a more proactive stance from the US government in protecting its technological advancements against foreign appropriation.
In addition to distillation, broader tactics employed by these foreign operations have raised alarms. Michael Kratsios, a key science and technology advisor to the government, indicated that Chinese entities are utilizing “tens of thousands of proxy accounts to evade detection” and employing jailbreaking techniques to expose proprietary information. This multifaceted approach has prompted the US to explore measures to hold foreign actors accountable for their actions.
In response to the accusations, the Chinese embassy in Washington dismissed the claims as “pure slander,” asserting that “China attaches great importance to the protection of intellectual property rights.” This denial highlights the ongoing tensions between the two nations regarding intellectual property and technology transfer issues.
Kratsios, who has served as a top advisor on science and technology under previous administrations, is part of the President’s Council of Advisors on Science and Technology (PCAST), which has recently been reconstituted to include prominent business leaders such as Mark Zuckerberg of Meta and Lisa Su of AMD. Given the memo’s origin and Kratsios’ involvement, it is likely to carry considerable weight in the ongoing dialogue surrounding AI and national security.
As the urgency to address these challenges intensifies, the implications for the US AI industry could be profound. With the recent launch of products like DeepSeek-V4, which offers capabilities like “cost-effective 1M context length,” there is a growing concern among American firms regarding the potential for corporate espionage and technology theft. These developments could reshape the landscape of competition in the global AI market.
Going forward, the US is not only looking to safeguard its technological assets but also to establish a framework that could deter foreign entities from engaging in such activities. As the discourse on AI security continues to evolve, American companies are likely to remain vigilant, seeking to shore up defenses against any vulnerabilities that could be exploited by foreign competitors.
See also
POET Technologies Stock Soars 31% Amid AI Optics Demand and $430M Funding Boost
Hippocratic AI Partners with Cincinnati Children’s and UNC Health to Enhance Patient Care Access
Ghana’s President Mahama Announces $250M Investment in National AI Computing Centre
NEC Partners with Anthropic to Create AI-Native Engineering for 30,000 Employees in Japan
DeepRoute.ai Equips 300,000 Vehicles in China, Aims for 1 Million More This Year
