China Rebuts 'AI Slowdown' Debate: 'Ruthlessly Crush' and 'Atomic Bomb' Warnings Revealed

A core developer at China's DeepSeek has directly challenged the argument from the U.S. AI industry that the pace of artificial intelligence development should be intentionally slowed down.

He argued, "In a situation where everyone is obsessed with self-destruction, I have no choice but to participate in a brutal arms race."

Liu Chengyu, a developer at Chinese AI startup DeepSeek, posted a long essay on his WeChat account on Monday (September 14).

Liu stated that even if he were to abandon AI model training or deliberately slow down development, models from other companies would continue to advance and eventually "ruthlessly crush" him.

He dismissed the idea that one side slowing down its pace in the AI development race is practically impossible.

In particular, he directly named and strongly criticized U.S. AI companies Anthropic and OpenAI.

First, Liu argued that as AI advances, it could lead to a devastated dystopia, claiming, "While a handful of tech companies control most of the resources and an ultra-minority uses the most advanced AI and various scientific technologies, most people will only use very weak AI."

He added, "Class mobility will also become increasingly difficult."

For this reason, he argued that cutting-edge AI should be supplied to everyone in an open and affordable manner, directly criticizing, "I don't believe Anthropic or OpenAI will do that. Especially if Anthropic takes control of cutting-edge AI or Artificial General Intelligence (AGI), to exaggerate somewhat, the severity would be equivalent to Hitler getting his hands on atomic bomb technology before the Allies."

This statement is drawing attention as a public rebuttal to the "AI slowdown" debate recently ignited within the U.S. AI industry.

Anthropic CEO Dario Amodei recently argued that a so-called "recursive self-improvement," where AI helps develop the next generation of AI on its own, is emerging, and asserted that "the pace of performance improvement in AI models must be slowed down."

OpenAI's Sam Altman and Elon Musk have also publicly agreed with this awareness of the issue.

The next day, Liu clarified that the post was merely his personal view and did not represent DeepSeek's official stance.

(Reported by Kim Minjeong | Video by Lee Eui-seon | Design by Lee Sumin | Produced by SBS Digital News)
※ Please note: This article was translated by AI and may contain errors.