Key Takeaways
- American AI companies OpenAI and Google delivered services to Singapore operations of Alibaba, Baidu, and Tencent—firms identified by the Pentagon as having military connections.
- These transactions remain lawful under existing U.S. regulations, which don’t prohibit AI service provision to Chinese entities operating beyond mainland borders.
- OpenAI terminated certain Alibaba-connected accounts following concerns about “distillation”—the practice of leveraging AI responses to develop competing models.
- Google acknowledges that location-based restrictions alone prove insufficient against determined efforts to circumvent access controls.
- Anthropic has adopted more stringent policies, refusing service to all Chinese-affiliated organizations while advocating for comprehensive U.S. export restrictions.
Leading American artificial intelligence companies OpenAI and Google have been delivering their cutting-edge AI model access to Singapore-based operations of three prominent Chinese technology corporations—Alibaba, Baidu, and Tencent—despite these firms’ parent entities being featured on a Pentagon register that connects them with China’s military apparatus.
The Financial Times broke this story on July 10, 2026.
Understanding the Pentagon’s Designation
The register in question is formally designated as the 1260H list. This catalog identifies corporations that U.S. authorities consider to have connections with China’s armed forces.
Inclusion on this roster doesn’t inherently prohibit companies from purchasing American AI capabilities. Present U.S. regulations contain no provisions preventing Chinese-based organizations from obtaining sophisticated AI model access when operating outside mainland Chinese territory, which explains the legality of these transactions.
Both OpenAI and Google verified to the Financial Times that they had furnished AI capabilities to these Chinese corporations’ Singapore-registered affiliates.
OpenAI’s Actions and the Distillation Challenge
OpenAI terminated service for certain Alibaba-connected accounts during the previous month. The organization stated it discovered potential “distillation” activities raising concerns.
Distillation describes a technique whereby developers exploit outputs from sophisticated AI systems to enhance and train their proprietary competitive platforms. OpenAI indicated it forwarded information about these activities to federal authorities.
OpenAI maintains it prevents direct mainland China access while permitting select Chinese-owned enterprises to utilize its offerings in jurisdictions where it considers enforcement mechanisms effective.
Google indicated its AI capabilities remain accessible in territories including Singapore and Hong Kong, governed by its terms of service. However, Google conceded that geographic limitations alone cannot prevent determined users from bypassing such restrictions.
Anthropic Implements Stricter Measures
Anthropic has implemented more restrictive policies than competing organizations. This company has prohibited Chinese corporations and their international subsidiaries from obtaining access to its advanced AI systems.
Anthropic has additionally alleged that Alibaba employed thousands of fraudulent accounts to extract information from its Claude AI platforms.
The organization is advocating for U.S. authorities to establish comprehensive export limitations on AI technology, mirroring existing constraints on sophisticated semiconductor shipments.
This disclosure has reignited discussions in Washington regarding whether AI export regulations have maintained pace with semiconductor restriction frameworks. Legislative officials and policy analysts are demanding strengthened guidelines governing frontier AI model access.
Alphabet stock showed minimal movement in premarket activity following the disclosure. Alibaba’s American-listed shares similarly remained unchanged.
