OpenAI, Google Served AI to Blacklisted Chinese Firms' Units

U.S. AI Leaders Supplied Services to Overseas Affiliates

OpenAI and Google have confirmed providing advanced artificial intelligence services to Singapore-based subsidiaries of Alibaba, Baidu, and Tencent, even as the parent companies appear on the Pentagon's list of businesses allegedly linked to China's military, according to details first reported by the Financial Times.

The transactions remained within current U.S. legal boundaries because existing regulations do not broadly prohibit Chinese-headquartered companies from accessing advanced AI models outside mainland China. However, the disclosures have amplified calls in Washington for stricter export controls on cutting-edge AI software comparable to existing restrictions on semiconductor technology.

Why It Matters

The gap between Pentagon blacklisting and AI export rules reveals a significant policy blind spot as Washington attempts to limit China's access to frontier technology. Unlike semiconductor equipment, which faces strict export controls, AI models can be accessed through cloud APIs from anywhere, making geographical restrictions difficult to enforce. This case demonstrates how Chinese companies can legally obtain advanced AI capabilities through overseas subsidiaries, potentially undermining broader U.S. technology containment efforts.

OpenAI Suspended Access After Detecting Model Distillation

OpenAI suspended API access last month for users connected to Alibaba after detecting possible "distillation"—a technique where developers use outputs from advanced AI systems to train competing models. The company notified the U.S. government following the discovery.

OpenAI stated it blocks direct access to its models from mainland China but permits certain Chinese-owned businesses to use its services in jurisdictions where appropriate safeguards can be applied. Google similarly confirmed its AI products remain available in markets including Singapore and Hong Kong under existing usage policies, while acknowledging that geographical restrictions cannot completely prevent sophisticated users from circumventing controls.

Anthropic Adopts Stricter Stance

Anthropic has taken a harder line by prohibiting Chinese companies and their overseas entities from accessing its frontier AI models entirely. The company has advocated for broader U.S. export controls designed to prevent advanced American AI technology from supporting the development of competing systems in China.

The Pentagon's 1260H list identifies businesses the U.S. government alleges have military ties to China. While inclusion on this list carries reputational and some procurement restrictions, it does not automatically trigger comprehensive technology export bans.

Shares of Alphabet and U.S.-listed Alibaba traded broadly flat following the disclosure.

These details were first reported by the Financial Times.