Anthropic published a new pape on Friday that calls for the U.S. to make sure China’s access to advanced chips is restricted, in the same week when President Donald Trump is in Beijing on a high-stakes state visit to China. #🇨🇳 ChinA.I. 🤖🧠🦾🤖
In the paper, entitled “2028: Two Scenarios for Global AI Leadership,” Anthropic said if the U.S. loosens its restrictions on China’s access to compute, China could become more competitive in #Artificial Intelligence development, resulting in a “destabilizing neck-and-neck race” two years from now.
“AI labs in China have world-class talent. It is compute constraints that limit their ability to keep up,” the U.S. company said.
Anthropic has been increasingly vocal about its view that China’s advancements in AI pose a threat to the U.S. and its allies. However, on social media sites such as X, some users criticized Anthropic’s new paper, saying that the company is using nationalism to its advantage in its attempt to fend off challenges from competitive Chinese open-source AI models. The paper comes as Chinese open-source AI developers, such as DeepSeek, have recently rolled out new models that offer more affordable alternatives to Anthropic’s proprietary models.
In the global AI race, it is “essential that the US and its allies stay ahead of authoritarian governments like the Chinese Communist Party,” Anthropic said in the paper. The prospect of the Chinese Communist Party leading in AI is “among the greatest threats to a successful transition,” it said.
In February, Anthropic accused multiple Chinese companies of using Claude to train their models, a technique known in the industry as “distillation.” DeepSeek, Moonshot AI and MiniMax had been carrying out “industrial-scale campaigns,” setting up fraudulent accounts to “illicitly extract Claude’s capabilities to improve their own models,” Anthropic said. The three Chinese companies didn’t respond to these allegations. #ScholER
anthropic.com/research/20..
