{"text":[[{"start":8.55,"text":"OpenAI and Google are selling their advanced AI models to Chinese tech giants blacklisted by the Pentagon, exposing a gap in Washington’s efforts to slow Beijing’s AI development."}],[{"start":21.200000000000003,"text":"The US companies confirmed to the FT that they have been supplying AI services to Singapore-based subsidiaries of Alibaba, Baidu and Tencent, which the US government has accused of working with China’s military. "}],[{"start":33.900000000000006,"text":"After being contacted by the FT this week, OpenAI said it had last month suspended Alibaba-affiliated users’ access to its API, the software interface that lets developers remotely access AI models, over concerns about illicit use. Although the sales are legal, they have reignited calls for tighter US regulation of AI models similar to its restrictions on exports of the chips used to train powerful models."}],[{"start":58.95,"text":"The US government has moved to control access to individual advanced AI models such as Anthropic’s Mythos and Fable and OpenAI’s GPT-5.6. But it has not banned the use of cutting-edge AI software more broadly by Chinese-headquartered entities, even those on the so-called 1260H list, a congressionally mandated blacklist that includes Chinese companies with alleged ties to the People’s Liberation Army."}],[{"start":84.80000000000001,"text":"“The [Trump] 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,” said Chris McGuire, a technology and security expert at the Council on Foreign Relations. OpenAI’s recent block of Alibaba users followed suspected “distillation”, in which developers use the outputs of AI models to improve rival systems. The activity was flagged to the US government, a spokesperson said. "}],[{"start":114.4,"text":"OpenAI said it does not allow its models to be accessed in China but confirmed that it allows “some companies” with Chinese ownership or headquarters to use its tools for operations in “countries where we can enforce safeguards and monitor for distillation”.It added: “we would rather see more of the world using AI shaped by democratic values than AI controlled by autocratic governments”. They added that “we don’t think nationality alone should decide access”."}],[{"start":140.20000000000002,"text":"Google said its AI services are available in Hong Kong and Singapore, subject to its usage policies, including the banning of distillation. But it added that geographic sales restrictions were not enough to mitigate distillation risk, as they could be easily circumvented by sophisticated attackers. "}],[{"start":158.75000000000003,"text":"Alibaba did not immediately respond to requests for comment. Last month, the company asked a US court to order the Pentagon to remove it from the 1260H blacklist, describing the designation as “arbitrary and capricious”.Baidu declined to comment. Tencent did not respond to request for comment on how it was using the US AI models. "}],[{"start":180.80000000000004,"text":"By contrast, Anthropic has banned Chinese companies and foreign entities owned by them from using its advanced models. "}],[{"start":188.20000000000005,"text":"The position has been difficult to enforce. Last week, it moved to shut loopholes that have allowed some Chinese companies to circumvent the AI group’s stringent restrictions on unauthorised use in the country.“Frontier AI models should not be served to China-headquartered companies anywhere in the world,” added McGuire, who worked on export controls as a senior White House and state department official during the Biden administration.Anthropic has previously accused Chinese AI labs DeepSeek, Moonshot and MiniMax of distillation. Last month, it warned in a letter to Congress that Chinese ecommerce firm Alibaba used 25,000 fraudulent accounts to generate more than 28.8mn exchanges with Claude, which it said was a violation of its terms of service. Alibaba did not respond to a request for comment about the accusations from Anthropic at the time. Joe Khawam, a specialist in AI policy and national security law at the Law Reform Institute, wrote earlier this year that the US government needed to act via export controls and related measures to avoid “Chinese laboratories that systematically extract frontier capabilities without bearing the compute, engineering, and safety costs” of American AI companies. "}],[{"start":265.65000000000003,"text":"He warned that this resulted in an erosion of “the economic foundation of US frontier AI leadership”."}],[{"start":273.35,"text":"Additional reporting by Eleanor Olcott "}],[{"start":283.25000000000006,"text":""}]],"url":"https://audio.ftcn.net.cn/album/a_1783680801_5984.mp3"}