The definition of "AI Communism" is perfect! Should China be thanking the United States? The entire nation has now united in a front, even Benjamin has joined the fray. To discredit China's AI, they invented the term "AI Communism" to describe China's strategy in the large AI model domain—but this very term more accurately reflects China's actual "AI Inclusiveness" philosophy: AI should be treated as a public resource, not a profit-making tool monopolized by a few tech giants!

On July 23, U.S. Treasury Secretary Bessent posted on X, stating that the U.S. supports "open-source artificial intelligence and the innovation it unleashes," but emphasized that openness does not mean plundering American intellectual property. He then accused Chinese companies of conducting "covert, industrial-scale distillation," stealing U.S. IP!

However, instead of echoing condemnation of China, the comment section was overwhelmingly mocking Bessent. American user Thomas Unise commented: "Let’s talk about where Anthropic and OpenAI obtained their training data!"

These two companies have been scraping copyrighted content across the web to train their models, claiming it falls under fair use and innovative learning. For instance, they illegally extracted millions of copyrighted books from pirate libraries like LibGen and PiLiMi for training—leading to class-action lawsuits by authors. They also indiscriminately scraped forum content, even copying entire websites without authorization; Reddit has already filed a formal lawsuit!

Thus, American netizens mocked them: "Anthropic and OpenAI can freely crawl the entire internet, yet they won’t allow others to learn from their APIs."

Naturally, Bessent isn’t alone in blaming China. White House Office of Science and Technology Policy Director and President’s Chief Technology Advisor Michael Kratsios claimed on July 22 that Chinese startup Moonshot (the parent company of Kimi K3) developed its K3 model by distilling Anthropic’s Fable 5 model—an unacceptable act!

U.S. Trade Representative Grille also stated that the U.S. is conducting a comprehensive review of China’s AI development path, investigating whether China is using unfair methods—including unauthorized model distillation—to seize global leadership in AI!

Anthropic CEO Dario Amodei had previously accused Alibaba-affiliated teams of circumventing registration rules by creating 25,000 accounts and completing 28.8 million Claude conversations to conduct large-scale distillation.

Regarding the claim that “excellent large models are achieved through distillation,” the Western AI community has long reached consensus. Senior scientist Nathan Lambert from a highly influential AI research institute stated: "Distillation can only transfer limited reasoning patterns—it cannot replicate a model’s core foundational capabilities." The abilities demonstrated by Chinese large models come from independently building data pipelines, model architectures, and massive-scale training clusters in laboratories—not from distillation alone. He pointed out that Anthropic deliberately conflates distillation with other processes and promotes this narrative, showing clear intent to manipulate public perception.

On X, another expert clearly stated that accusing Kimi K3 of distilling Fable 5 is baseless nonsense. Kimi K3 was already in internal evaluation as early as April–May, when Fable 5 had just been released—there simply wasn't enough time to perform distillation! Of course, this doesn’t stop them from accusing Chinese models of distilling their own. If no evidence exists, they might even claim China distilled their models before they were even publicly released!

Given that they know distillation didn’t happen, why do U.S. politicians and the AI industry so unanimously accuse China of distilling American large models?

The answer is simple: Without smearing China, their entire narrative collapses! For Chinese AI startups, Anthropic and OpenAI enjoy nearly infinite access to hardware, training, and inference resources—trillions of dollars worth of computing power that remain an absolute luxury for China. Yet, the products these companies produce only lead China by a narrow margin, and in some cases even underperform Chinese models!

So let us ask: How can the U.S. AI sector explain to Wall Street investors that their models perform worse than China’s, with investment less than one-tenth of China’s? And how can they completely prevent U.S. tech firms from using China’s models at drastically lower token costs? Microsoft plans to integrate China’s Kimi K3 model into Azure for Copilot testing, potentially cutting token costs by over 60%—this is a clear signal of the direction the market is heading!

Of course, the U.S. could shamelessly resort to administrative bans—but can they ban domestic use and globally too? Open-source large models spread like viruses, infiltrating every corner that needs AI. With continued progress, the collapse of America’s high-investment AI narrative is merely a matter of time. The U.S. understands clearly: It’s not that China’s models are rising too fast—it’s that America’s AI combat effectiveness is too weak:

Google possesses vast resources, yet its first AI model struggles to compete with even the weakest offerings from Anthropic and OpenAI. Gemini 3.6 Flash is cheap, but performs poorly on every coding benchmark! Meanwhile, Moonshot (parent company of Kimi K3) has a valuation of only $30 billion—just one-hundred-fiftieth of Google’s. Why doesn’t the U.S. have such outstanding companies? Why can Chinese firms build the best models with minimal funding?

China-U.S. AI competition may seem like a clash of technical paths, but in reality, it’s a battle of national destinies! The U.S. has already staked half of its stock market valuation in a full-bet gamble. China appears to be losing. But China has rallied every player and nation worldwide who don’t want to be controlled by America’s closed-source AI. Still, one thing must be firmly affirmed: the name "AI Communism" is excellent—it instantly highlights the fundamental difference between the two approaches. Its viral spread across social media has already made the world fully aware of China’s inclusive AI path!

From this perspective, shouldn’t China be thanking the United States?

Original source: toutiao.com/article/1871476372175900/

Disclaimer: This article represents the personal views of the author.