Anthropic Urges Swift US Action to Maintain AI Dominance over China
- Anthropic has issued a call for the United States to expedite efforts to secure a one- to two-year lead in artificial intelligence over China.
- The firm posits that the U.S. possesses advantages in computational power and has urged policymakers to eliminate loopholes facilitating chip transfers to China.
- Moreover, Anthropic emphasized the necessity for the U.S. to “prevent squandering our competitive edge.”
Anthropic contends that the United States can solidify a one- to two-year AI advantage against China, provided swift action is taken.
In an extensive post published on Thursday, the company outlined two potential trajectories for the AI landscape by 2028: one scenario entails the U.S. restricting Chinese access to American AI computation resources, while the alternative does not.
According to Anthropic, China is rapidly bridging the AI divide with the U.S. through lax regulations on chip exports and the practice of distillation attacks, which involve leveraging an advanced AI model to enhance a smaller “student” model.
The organization stated, “If the U.S. and its allies take decisive actions now, it may be feasible to secure a 12- to 24-month lead in cutting-edge capabilities.”
Anthropic cautioned that “the opportunity to solidify that advantage may not persist indefinitely.”
Securing a lead is critical for ensuring the safety of AI, the firm added, highlighting that a “neck-and-neck competition between American and Chinese AI laboratories could complicate safety and governance measures led by both industry and government.”
Intensifying competition exerts pressure on AI laboratories in both nations to expedite model releases without thoroughly assessing safety protocols, it noted.
Anthropic advocated for the implementation of policy reforms aimed at enhancing chip export controls, augmenting enforcement budgets, and thwarting distillation attacks perpetrated by Chinese AI enterprises.
“Our historical achievements necessitate that our current objective is to avoid wasting our advantage: we must choose not to facilitate the Chinese Communist Party’s catch-up efforts,” the firm remarked.
In February, Anthropic disclosed that three of China’s leading AI firms—DeepSeek, MiniMax, and Moonshot AI—were “illicitly” using Claude for the enhancement of their own models.
The Biden administration initially enacted export controls on American chips to China in 2022, and the Trump administration built upon these regulations, prohibiting Nvidia and AMD from supplying chips to China.
However, last August, it partially reversed the decision, allowing Nvidia to market its H200 chips contingent on a 25% levy to the U.S. government.
Moreover, clandestine activities by Chinese entities seeking American chip access persist. In December, the U.S. indicted several individuals for attempting to smuggle Nvidia’s cutting-edge chips by mislabeling them as “SANDKYAN” chips.
Despite Anthropic’s claims regarding China’s AI advancements, former ByteDance engineer Zhang Chi asserted in April that China is, in fact, falling further behind.
Zhang, now a research scientist and assistant professor at Peking University, noted in a podcast interview that Chinese AI suffers from inadequate high-quality data for model training, as well as limited access to sophisticated chips.
Anthropic’s assertions coincided with a visit by President Donald Trump to China, marking the first occasion a U.S. president has visited since Trump’s own trip in 2017.
Accompanying him were prominent U.S. business figures, including Tesla’s CEO Elon Musk, Apple’s CEO Tim Cook, and Nvidia’s CEO Jensen Huang.
Source link: Aol.com.
