Google and OpenAI Provide AI Access to Chinese Tech Giants via Singapore Subsidiaries
Translated & summarized from Maariv by baba
Google and OpenAI have provided AI services to Singapore subsidiaries of blacklisted Chinese tech firms, exploiting a legal loophole that circumvents U.S. restrictions. This has raised concerns in Washington about tightening AI regulations to prevent technology transfer to China. OpenAI recently blocked Alibaba-linked users and reported them to U.S. authorities, while Google and Anthropic have varying enforcement policies regarding Chinese access.
The story in 6 lines · by baba
- Google and OpenAI supplied AI services to Singapore subsidiaries of blacklisted Chinese tech giants.
- This access exploits a legal loophole bypassing U.S. efforts to restrict China's tech development.
- OpenAI blocked Alibaba-linked users last month and reported suspicious activity to U.S. officials.
- Google admits geographic blocks can't fully prevent AI model refinement by Chinese firms.
- Anthropic bans Chinese companies entirely from using its AI models despite enforcement challenges.
- The situation exposes gaps in U.S. policy on restricting AI technology transfers to China.
A Financial Times investigation revealed that Google and OpenAI have supplied artificial intelligence services to Singapore-based subsidiaries of major Chinese technology companies, including Alibaba, Baidu, and Tencent. These companies are blacklisted by the U.S. Pentagon for alleged cooperation with the Chinese military. The access is enabled through a legal loophole allowing American firms to work with subsidiaries located in Singapore, effectively bypassing Washington's efforts to slow China's technological advancements.
While these transactions are legal since they do not involve direct dealings with the Chinese headquarters, they have reignited calls in Washington for stricter AI regulations similar to existing export controls on semiconductor chips. The concern centers on a process called "refinement," where companies use outputs from advanced AI models to enhance competing systems.
OpenAI confirmed that it recently blocked access to its development interface for users linked to Alibaba amid suspicions of unauthorized use and reported this to the U.S. government. An OpenAI spokesperson emphasized that although their models are not available within China, they permit Chinese-owned corporations to use their tools in regions where usage policies can be enforced and suspicious activity monitored. The spokesperson added that OpenAI prefers AI development aligned with democratic values rather than control by authoritarian regimes.
Google acknowledged that its services are accessible in Singapore and Hong Kong under similar restrictions but admitted that geographic blocks alone cannot fully prevent sophisticated refinement processes. In contrast, AI company Anthropic has taken a stricter stance by completely banning Chinese companies and their foreign-owned affiliates from using its models, despite enforcement challenges.
This complex situation highlights gaps in U.S. policy, which has yet to impose a comprehensive ban on blacklisted Chinese firms from accessing advanced AI software through international subsidiaries, potentially undermining efforts to maintain America's technological edge.
Mentioned
The same event, reported separately by each outlet. Open a few to compare what different newsrooms emphasize — and what they leave out.
Centre 1Haredi 1
