OpenAI and Google have provided their leading artificial intelligence models to Chinese tech giants included in the Pentagon's 'blacklist,' according to the Financial Times.
The American companies confirmed to FT that they have offered AI services to subsidiaries of Alibaba, Baidu, and Tencent registered in Singapore, which U.S. authorities accuse of collaborating with China's armed forces.
OpenAI stated that it had suspended access to its API (application programming interface) for users associated with Alibaba last month, which allows developers to remotely use the company's AI models. The article notes that this decision was driven by concerns over potential illegal use.
The company also told FT that it does not provide access to its models within China but confirmed that it allows 'certain companies' with Chinese owners or headquarters to use its tools in countries 'where we can ensure the preservation of security measures and monitor cases of model distillation.'
OpenAI added that they would like as many countries as possible to use AI shaped by democratic values rather than AI controlled by authoritarian regimes. The company also emphasized that they 'do not believe that only national affiliation should determine the right to use technologies.'
In turn, Google informed FT that its AI services are available in Hong Kong and Singapore, provided that usage rules are followed. The company also noted that geographical restrictions alone are not sufficient to reduce the risk of model distillation, as experienced abusers can easily circumvent them.
Although, as the publication notes, such transactions are currently legal, they have reignited calls in the U.S. to strengthen controls over the export of AI models, similar to existing restrictions on chip exports.
The U.S. government has already taken steps to control access to certain leading AI models, including Anthropic's Mythos and Fable models, as well as OpenAI's GPT-5.6 model.
However, according to the article, Washington has not yet fully banned the use of leading AI software by Chinese organizations, even if they are included in the so-called 1260H list approved by the U.S. Congress, which includes companies that, according to U.S. authorities, are linked to the People's Liberation Army of China.
