Reviewing China’s AI big models, the US is once again “slinging accusations,” experts say: Clearly unreasonable and not credible.

[Global Times Report by Chen Zishao] According to a report by Reuters on the 9th, U.S. officials have been secretly evaluating whether China’s large AI models are outputting content in line with China’s official stance. U.S. officials revealed that they might publicly evaluate the results to issue a warning to China about the deployment of so-called “ideologically oriented AI tools.” Liang Huaixin, a researcher at the National Institute for National Security and Governance at the University of International Business and Economics, told Global Times on the 10th that the U.S.’s review of China’s AI models is a typical example of shaping technology positions through political ideology, “which is obviously unreasonable and unfounded.”
In a memorandum seen by Reuters, it mentioned that U.S. State Department and Department of Commerce officials have been testing Chinese company-developed AI models recently, scoring them based on whether the models answer questions and how closely their answers align with the official political stance. Testing results showed that Chinese AI tools tend to align more closely with official stances than similar products from the United States.
Regarding this, Reuters stated that whether AI model developers can influence the ideological orientation of their chatbots has become a core concern, not limited to Chinese AI models. Elon Musk, a billionaire in the United States, recently modified his Grok model after modifying his company’s Grok model, which began to express support for Hitler and attacks against Jews. On the 8th, Musk issued a statement saying that Grok was “actively removing inappropriate content.”
Liang Huaixin told that the U.S.’s review is unfounded. The development of any AI model is carried out under normal legal supervision in the country, requiring compliance with relevant national laws and regulations. When China’s models operate in other countries, they will definitely comply with local laws. This issue should be considered from a legal perspective rather than a political one, otherwise it could lead to issues of labeling. He believes that what the U.S. should focus on now is not trying to catch so-called “China’s gripes” or generalizing security measures against China but should instead work hand in hand with China and other countries to promote global governance of AI models, “that is the right path.”