Summary：**Alarming report: OpenAI and Google sell frontier AI to blacklisted Chinese giants****Introduction

referrerpolicy="no-referrer"
style="max-width:100%;height:auto;display:block;margin:0 auto;">

**Alarming report: OpenAI and Google sell frontier AI to blacklisted Chinese giants**

**Introduction**
A recent investigation reveals that three Chinese technology firms listed on the U.S. Department of Defense’s Entity List continue to acquire cutting‑edge artificial intelligence models from OpenAI and Google, despite formal export restrictions. The loophole hinges on routing purchases through third‑party jurisdictions where U.S. sanctions do not apply, allowing the blacklisted companies to access the same frontier AI tools that power advanced research in the United States.

**Key Developments**
The investigation, conducted by a coalition of trade analysts and cybersecurity experts, identified that Baidu, Alibaba, and Tencent have each secured access to generative language models and vision systems through subsidiaries located in Singapore, the United Arab Emirates, and Luxembourg. These entities purchase API credits or cloud‑based AI services directly from OpenAI’s and Google’s platforms, then transfer the capabilities back to the parent companies in China.

U.S. officials confirmed that the Entity List prohibits direct transfers of “dual‑use” technology to the named firms, but current regulations do not explicitly cover indirect acquisition via foreign intermediaries. In response, the Bureau of Industry and Security (BIS) issued a statement acknowledging the gap and pledging a review of existing export control frameworks to close the loophole.

**Industry Analysis**
Industry analysts warn that the practice undermines the strategic intent of the blacklist, which aims to prevent adversarial nations from enhancing military‑grade AI capabilities. “Frontier models are not just consumer products; they are foundational tools for autonomous systems, surveillance, and cyber operations,” said Dr. Maya Patel, a senior fellow at the Center for Strategic and International Studies. “When blacklisted firms can obtain these models through offshore channels, the effectiveness of sanctions is significantly diluted.”

Moreover, the trend raises concerns about intellectual property leakage. Although the API agreements prohibit redistribution, enforcement across borders remains challenging, and there is evidence that fine‑tuned versions of the models are being deployed in domestic Chinese AI labs for applications ranging from facial recognition to natural‑language processing for defense contractors.

**Future Outlook**
Policymakers are expected to tighten regulations governing cloud‑based