A DeepSeek engineer has sharply criticised the prospect of US companies controlling the world’s most advanced artificial intelligence, comparing potential dominance by Anthropic to Nazi Germany acquiring nuclear weapons before the Allies.
Liu Shengyu, a machine-learning systems engineer working on DeepSeek’s V4.1 model, made the remarks in a Chinese-language blog post in which he advocated making cutting-edge AI widely and affordably available through open models.
“I believe cutting-edge intelligence should be made available to everyone in an open and affordable manner,” Liu wrote.
He said he did not trust Anthropic or OpenAI to pursue that approach, contrasting them with DeepSeek’s open-source strategy.
“I do not trust Anthropic or OpenAI to do this; specifically, I dread the prospect of Anthropic controlling the most advanced AI or AGI,” Liu said. “To put it dramatically, the consequences would be no less severe than Hitler acquiring atomic bomb technology before the Allies.”
AGI, or artificial general intelligence, broadly describes hypothetical AI capable of matching or exceeding human-level intellectual abilities across a wide range of tasks.
Liu also argued that increasingly powerful AI could have major consequences for economic opportunity and social mobility.
As AI advances, he said, moving upward in society becomes “increasingly difficult: you need the most powerful AI to break through class barriers.”
His comments come as prominent figures within the US AI industry debate whether frontier development should be slowed because of potentially catastrophic risks.
Anthropic CEO Dario Amodei called over the weekend for slower and more closely monitored AI development. He proposed placing independent third-party evaluators inside leading AI laboratories, while arguing that safeguards should be designed without “sacrificing commercial advantage.”
Former Anthropic researcher Jacob Coxon resigned last week after accusing leading AI companies of failing to act responsibly. Anthropic alignment-science lead Evan Hubinger subsequently said he personally estimated the probability of AI causing human extinction at greater than 10 per cent within the next decade.
Amodei’s proposals have also drawn criticism in China. The independent Dongbuyaqiao Research Institute suggested on Tuesday that his position could represent “an attempt to gain greater control over industry rules before they are fully established.”
Chinese Foreign Ministry spokesperson Guo Jiakun said on Monday that “fearmongering, confrontation and vicious competition” would obstruct efforts to establish effective global AI governance.
Meanwhile, a commentary in the state-run Global Times characterised Amodei’s proposals as a “silent AI Cold War,” arguing that they could help preserve US technological dominance while limiting China’s role in shaping global AI rules.
US President Donald Trump has opposed calls to slow AI development, arguing that restrictions on American progress could allow China to gain a competitive advantage.