China says Anthropic has a disease, American AI companies can't dictate rules for rest of the world

China claims that Anthropic has a disease and wants the US to not let its AI companies dictate rules for the world. This comes just days before President Xi Jinping's visit to Washington DC to meet US President Donald Trump.

In Short

• China takes dig at Anthropic, says it has a disease
• China argues American AI companies cannot dictate AI rules
• Beijing is set to have talks with the US on AI soon

Ahead of their high-profile meeting scheduled for later this month, China has lashed out at the US, and more specifically, American AI companies like Anthropic for trying to dictate AI rules for the rest of the world. China in fact is setting some new terms and conditions before its premier, President Xi Jinping, meets with his US counterpart, President Donald Trump later this month.

In a social media post titled, “Anthropic Has Contracted the American Disease,” Yuyuantantian, a social media account affiliated with Chinese state broadcaster CCTV, said the US must prove that American AI companies need to comply with the same safety, disclosure, and audit rules before any “substantive” discussions with China can take place. As per reports, this account is frequently used to signal Beijing’s official policy thinking.

Officials from the US and China are expected to hold talks on AI before Xi’s visit, although no date has been announced for that dialogue.

China takes dig at Dario Amodei and Anthropic

In the post, the account mentions that Anthropic CEO Dario Amodei believes that “if powerful AI develops in the right direction, the world will become ‘stunningly beautiful.’”

But the post argues, to realise the utopia he describes, “Anthropic needs enormous investment to build the world’s most advanced models and reach users throughout society.” The account also accuses the AI startup of using user data to train its AI models.

“Anthropic treats every interaction and every line of code as part of its ability to train its models,” the post says. “Since all this is supposedly in pursuit of humanity’s ultimate ideal, then along the way, what does it matter if a little user data is collected, a little terminal access is taken over, or a little user consent is bypassed?”

The post further added that as Anthropic plans to go public later this year, potentially at over a trillion-dollar valuation, it now “scrambled for power and advantage,” showing that “it has long since ceased to be merely a technology company.”

Anthropic has accused Chinese AI labs of distilling its models in the past. Distillation refers to the process of training an AI model on a larger AI model.

China sets terms for AI talks with the US

The comments come as tensions rise over the AI race between the world’s two biggest economies. Underscoring that point, the account claimed that the way US models were developed, changed how a negotiation between the two countries could take place. “The problem is that America’s own frontier models have already developed in a distorted direction,” the post said. “This means the negotiation is not simply a technical dialogue from the start, but a continuation of all the earlier problems – the US is trying to turn the ‘safety boundaries’ it has drawn into the default rules for the entire world.”

“A clear distinction must be drawn between genuine security threats and mere technological competition,” the post said. “This line must be drawn jointly by all participating parties.” The post added that the US push for AI “control” was in effect aimed at making China accept an order partly defined by American companies.

The post concluded, “Only after the USs first proves that its security rules apply equally to its own model companies will it be possible for China and the United States to enter into substantive discussions.”

In recent months, the US has increased scrutiny over advanced AI models, with the likes of Anthropic’s Claude Mythos temporarily being restricted for use by non-Americans. Meanwhile, the White House has also considered banning open-weight AI models, including those made by Chinese companies, such as Kimi K3 and DeepSeek V4.

As per reports, Chinese officials are also increasingly concerned that Anthropic’s Mythos could be used against their country. It is believed that the US is working with engineers from Anthropic to potentially use Mythos for global cyber operations. On the other hand, Anthropic had blocked its services from Chinese-controlled companies last year, citing US national security concerns, although that does not appear to have fully stopped Chinese companies from using its products.

- Ends