DeepSeek Developer Compares Anthropic's AI Domination to Hitler Getting the Atomic Bomb First

A core developer at Chinese artificial intelligence (AI) startup DeepSeek has launched a scathing critique of Anthropic, emphasizing that a handful of companies must not monopolize the technology to prevent the perils that AI advancements could bring.

In a post uploaded to his WeChat account on September 14, DeepSeek developer Liu Chenghui wrote, "Humanity has historically never hesitated in the slightest when it comes to self-destruction. Even if I were to give up or intentionally delay and disrupt model training, models from other companies would continue to advance and ultimately crush me without mercy. In a situation where everyone is obsessed with self-destruction, I, too, have no choice but to participate in this brutal arms race."

He projected that depending on AI advancements, future society could either head toward communism—where productivity is maximized and people's living standards are significantly enhanced—or a world resembling the desolate dystopian video game "Cyberpunk 2077."

Liu pointed out, "In the latter (dystopia), a small number of tech companies would control most resources, and a tiny fraction of people would use the most advanced AI and various science and technology to approach 'mechanical ascension,' while most people would only be able to use very weak AI. Social mobility would become increasingly difficult."

He continued, "I still believe that cutting-edge AI should be supplied to everyone in an open and affordable manner. I do not believe that Anthropic or OpenAI will do that. Especially if Anthropic takes control of cutting-edge AI or AGI (Artificial General Intelligence—AI that can perform tasks with human-like capability), to put it somewhat exaggeratedly, the severity would be on par with Hitler getting his hands on atomic bomb technology before the Allied forces."

Liu's post originated from his reflections as a developer following the announcement of DeepSeek's new model v4.1, but the passage criticizing Anthropic drew widespread attention, garnering over 100,000 views within two days.

In particular, coinciding with Western AI industry figures such as Anthropic CEO Dario Amodei advocating for "AI safety brakes" and the necessity of regulating China, overseas social media platforms like X (formerly Twitter) and foreign media outlets also showed strong interest.

In a separate post the following day, September 15, Liu clarified that the original purpose of the post was not to discuss Anthropic as "unlikable," and that it merely reflected his personal views and did not represent the company's official stance.

He emphasized, "It is certain that AI will continuously grow stronger in the future, and physical AI, recursive self-improvement (RSI), and other technologies we do not yet know are waiting for us. I am optimistic about the capabilities of future AI."

(Photo: AP, Yonhap News)
※ Please note: This article was translated by AI and may contain errors.