OpenAI, Google Supplied AI Services to Pentagon-Blacklisted Chinese Firms Via Singapore Units: Report
OpenAI and Google have been supplying advanced AI services to Chinese technology companies blacklisted by the Pentagon, according to a Financial Times report. The services went to Singapore-based subsidiaries of Alibaba, Tencent and Baidu, and the arrangement is now raising fresh questions about how well US export controls actually work.
Alibaba, Baidu and Tencent were recently added to the Pentagon’s 1260H list, which names companies with alleged ties to China’s military. OpenAI and Google confirmed providing AI services to these companies’ Singapore-incorporated units anyway.
The reason it’s legal: the subsidiaries are registered in Singapore, so they can enter into agreements with US AI providers even when their parent companies are blacklisted. It’s the kind of structural gap that’s made AI export rules much harder to enforce than the existing controls on semiconductor exports.
Singapore’s position as a regional tech hub is part of what makes this possible. Multinational firms can operate there under Singapore’s legal framework while serving customers across Asia, which has turned the country into a key access point for AI services and a growing focus for US regulators.
Frontier AI isn’t just a commercial product anymore. It’s treated as a national security asset. The US restricts access to advanced AI from mainland China directly, but current rules don’t broadly stop overseas subsidiaries of Chinese companies from buying AI models from American firms.
OpenAI told the Financial Times it blocks direct access to its models from China, but allows certain Chinese-owned companies to use its services in jurisdictions where safeguards can be enforced. Google said its AI services stay available in markets including Singapore and Hong Kong, subject to its usage policies, and added that geographical restrictions alone aren’t enough to stop sophisticated users from getting around existing controls.
This comes shortly after OpenAI disclosed it had suspended API access for Alibaba-affiliated users last month over suspected “distillation” activity, where developers use outputs from advanced AI models to improve their own competing systems. OpenAI said it reported the activity to the US government.
Anthropic has gone further than either company, barring China-headquartered firms and all their foreign subsidiaries from its most advanced models. It’s also made distillation accusations of its own, against DeepSeek, Moonshot and MiniMax. In Congressional testimony in June, Anthropic alleged that Alibaba used tens of thousands of fraudulent accounts to generate more than 28 million interactions with Claude as part of a distillation effort.
Also Read:Unclaimed EPF Savings Worth Rs 9,330 Crore Parked In 31 Lakh Dormant Accounts
The broader US-China tech rivalry has moved well past hardware at this point. Since 2022, the US has tightened export controls on advanced AI chips, especially Nvidia’s, while expanding its blacklists of Chinese tech companies. The Biden administration introduced the initial semiconductor restrictions; the Trump administration kept tightening them, arguing the technology could strengthen China’s military and intelligence capabilities.
China has hit back by tightening its own export controls on gallium, germanium and rare earth elements, all critical to advanced manufacturing. Meanwhile, Chinese firms like DeepSeek, Alibaba, Baidu and Tencent have sped up development of their own large language models to cut reliance on US technology, backed by rising government investment in domestic AI and chip production.