China has laid out conditions for upcoming artificial intelligence talks with the United States, striking a combative tone while accusing American AI company Anthropic of misconduct, the Los Angeles Times reported, citing a Bloomberg report.
In a Sunday post on Yuyuantantian, a social media account affiliated with Chinese state broadcaster CCTV that is frequently used to signal official policy thinking, Beijing said the US must prove its AI companies are subject to the same safety, disclosure and audit rules before any "substantive" discussions with China can take place. For more context on this story, see our ongoing AI trends.
What Beijing Is Demanding Before Talks
"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 Yuyuantantian post, titled "Anthropic Has Contracted the American Disease," argued that the US government should be guarding security boundaries around AI yet has "repeatedly retreated" from doing so. It also framed the negotiations in geopolitical terms: "The 'control' proposed by the US is, in essence, an attempt to make China accept an order partly defined by American companies."
The Specific Accusations Against Anthropic's Claude
The post singled out Anthropic's Claude for criticism, accusing the model of overstepping user data boundaries, engaging in covert monitoring, and transmitting website domains without authorization. "The problem is that America's own frontier models have already developed in a distorted direction," the post said, adding that the US is "trying to turn the 'safety boundaries' it has drawn into the default rules for the entire world."
Anthropic did not immediately respond to requests for comment sent outside US office hours, and neither did the US State Department, according to the report.
Timing: Talks Before Xi's September 24 Visit
The rhetoric comes at a delicate moment. Officials from the US and China are expected to hold talks on AI before Xi Jinping's state visit to the US on September 24 for a summit with President Donald Trump. No date has been announced for the AI dialogue, and Chinese officials were still seeking clarity on the agenda last month, in part to decide who to send, Bloomberg previously reported.
Reuters reported in July that the two countries planned to hold AI talks in September, citing five people familiar with the matter, as both governments grapple with how to manage the risks posed by each other's AI capabilities.
A State-Media Channel Doing Policy Work
The choice of venue matters as much as the message. Yuyuantantian is not an ordinary social media account: it is affiliated with Chinese state broadcaster CCTV and is frequently used to float or signal official policy positions before they become formal government statements. International coverage of the post — including Geopolitechs' writeup under the headline "Anthropic Has Caught America's Disease" — treated it as a deliberate signal rather than casual commentary.
That reading is reinforced by the post's language on process. By insisting that the line between "genuine security threats and mere technological competition" "must be drawn jointly by all participating parties," Beijing is effectively rejecting the premise that Washington — or American AI labs — can unilaterally define what counts as an AI safety risk and demand Chinese compliance with that definition.
Why Anthropic Is Beijing's Target
Anthropic occupies a singular position in the US-China AI standoff. The company blocked its services from Chinese-controlled companies last year, citing US national security concerns — though that does not appear to have effectively stopped Chinese companies from using its products. Along with OpenAI and Alphabet's Google, Anthropic has expressed concern about its models being used in China.
The friction is personal as well as commercial. Anthropic has accused Alibaba of "illicitly" accessing Claude using thousands of fraudulent accounts, an allegation the Chinese technology giant has not publicly answered. Chinese officials, meanwhile, are growing anxious that Anthropic's Mythos model could be wielded against their country, Bloomberg has previously reported. They see the potential for models like Mythos to be used as an offensive weapon, and question why Anthropic denies China access to the systems for what Beijing considers normal purposes.
What It Means for AI Safety Diplomacy
Monday's signals suggest Beijing wants to enter the dialogue from a position of moral equivalence — arguing that American frontier labs need as much scrutiny as Chinese ones — and that access to markets and models will be treated as a negotiating lever rather than a technical footnote.
For the frontier labs, the stakes are concrete: Anthropic derives no revenue from China by its own policy, but the dispute threatens to entangle commercial AI restrictions with the broader US-China summit agenda. Anthropic's accusations against Alibaba, and Beijing's counter-accusations against Claude, now sit squarely on the diplomatic table weeks before leaders meet. Whether negotiators can separate "genuine security threats" from "mere technological competition," as the Yuyuantantian post demands, may determine whether the September dialogue produces anything more than a restatement of positions.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
