OpenAI and Google have been providing advanced artificial intelligence services to Singapore-based subsidiaries of Chinese technology companies blacklisted by the Pentagon, highlighting a gap in U.S. efforts to restrict Beijing's access to cutting-edge AI technology.
The companies confirmed to the Financial Times that they have supplied AI services to affiliates of Alibaba, Baidu, and Tencent, all of which have been placed on the Defense Department's 1260H blacklist over alleged ties to China's military, the newspaper reported Friday.
While the transactions are legal under current U.S. regulations, they have renewed calls for tighter export controls on advanced AI models.
OpenAI said it suspended Alibaba-affiliated users' access to its application programming interface, or API, last month after identifying suspected "distillation," a process in which developers use the output of one AI model to improve competing systems.
An OpenAI spokesperson said the company reported the suspected activity to the U.S. government.
OpenAI said it prohibits access to its models from within China but allows "some companies" with Chinese ownership or headquarters to use its AI tools in countries "where we can enforce safeguards and monitor for distillation."
"We would rather see more of the world using AI shaped by democratic values than AI controlled by autocratic governments," the company said, adding that "we don't think nationality alone should decide access."
Google said its AI services are available in Hong Kong and Singapore, subject to company policies that prohibit distillation and other misuse. The company added that geographic restrictions alone are insufficient because sophisticated users can circumvent them.
The disclosures have intensified debate over whether the Trump administration should impose broader export controls on advanced AI software, similar to restrictions already placed on the advanced semiconductor chips used to train leading AI models.
The U.S. government has taken steps to regulate access to certain frontier AI models, including Anthropic's Mythos and Fable models, and OpenAI's GPT-5.6. However, it has not broadly prohibited Chinese-headquartered companies, including those on the 1260H list, from using advanced AI services outside China.
"The administration says we need to beat China on AI all the time, but the problem is they haven't done anything on export control, which is the actual tool we have to slow China down," Chris McGuire, a technology and security expert at the Council on Foreign Relations, told the Financial Times.
Anthropic has taken a stricter approach, banning Chinese companies and foreign entities they own from using its advanced AI models. The company recently tightened its restrictions after identifying what it said were attempts by Chinese firms to bypass its safeguards.
Anthropic has previously accused Chinese AI companies, including DeepSeek, Moonshot, and MiniMax, of distillation. In a letter to Congress last month, the company alleged Alibaba used 25,000 fraudulent accounts to generate more than 28.8 million interactions with its Claude AI model in violation of its terms of service. Alibaba did not respond to requests for comment on those allegations.
Alibaba has challenged its inclusion on the Pentagon blacklist, asking a federal court last month to remove the designation, calling it "arbitrary and capricious." Baidu declined to comment, and Tencent did not respond to requests for comment on its use of U.S. AI models.
Joe Khawam, an AI policy and national security law specialist at the Law Reform Institute, argued earlier this year that stronger export controls are needed to prevent Chinese companies from extracting advanced AI capabilities without bearing the development costs borne by U.S. firms.
He warned that such practices could erode "the economic foundation of U.S. frontier AI leadership."