Anthropic Outlines 3 Measures to Keep US Ahead of China in AI

Anthropic has published a paper on the US-China competition in artificial intelligence (AI). It argues America can secure a 12-to-24-month lead over China with the right policy moves.

The United States and China are engaged in an intense competition for leadership in AI development. However, Anthropic stressed that the US and its allies must maintain their advantage over the Chinese Communist Party (CCP).

Anthropic Pushes Washington to Cement a 12-24 Month AI Lead

The paper noted that although export controls have helped preserve the United States’ current lead, the measures have not been sufficient. 

Follow us on X to get the latest news as it happens

The Chinese Communist Party (CCP) lacks the capacity to produce enough advanced chips domestically or legally acquire them from overseas.

Nonetheless, Chinese AI labs have remained competitive through two key workarounds. The first is illicit compute access, smuggling advanced AI chips into China and tapping into offshore data centers.

The second is illicit model access, which includes distillation attacks on US frontier systems and the use of those same models to speed up domestic AI research.

“But if the US and its allies act now to address both issues, it may be possible to lock in a 12-24 month lead in frontier capabilities. A lead that large by 2028 would be enormously advantageous,” Anthropic wrote. “But the window of opportunity to lock in that lead will not necessarily remain open for long.”

The paper cites a study that estimates that America would have 11 times as much compute than China if Washington tightened controls on the CCP’s access to US compute.

Thus, Anthropic outlined three policy actions to keep democracies ahead in AI.

• Tighten controls on chip smuggling, foreign access to data centers, and semiconductor manufacturing equipment to lower China’s compute ceiling.
• Legally clarify that distillation attacks are illegal and improve the US labs’ ability to detect and prevent them.
• Promote global adoption of trusted US AI hardware and models to deny the CCP a foothold in key markets.

Anthropic warned that AI could soon become powerful enough to “be used to repress citizens” on an unprecedented scale and potentially reshape the global balance of power. 

The company added that, because AI capabilities are advancing fast, there is only a narrow window to establish the rules and conditions governing the competition before those risks fully materialize.

Subscribe to our YouTube channel to watch leaders and journalists provide expert insights