The recent revelations about OpenAI and Google selling AI models to subsidiaries of Chinese tech behemoths like Alibaba, Baidu, and Tencent mark a significant breach in US export controls. These companies are on the Pentagon's 1260H blacklist due to suspected ties to the Chinese military, raising serious questions about the effectiveness of current regulations.

The loophole leveraged by these companies revolves around the subsidiaries located in jurisdictions like Singapore. This enables them to access advanced AI without triggering the stringent export controls that primarily focus on hardware, particularly advanced semiconductors. Such a structural oversight illustrates the challenges in adapting regulatory frameworks designed for physical goods to the digital realm.

OpenAI's acknowledgment of restricting access from China while allowing usage in 'enforceable jurisdictions' further complicates the narrative. Although OpenAI suspended API access for Alibaba-linked accounts due to signs of 'distillation' the process of using a robust AI model’s outputs to train a cheaper substitute the core structural issue remains unresolved. This scenario highlights the inadequacy of existing controls that were crafted for a bygone era, failing to account for the ubiquitous nature of software delivery.

In June 2026, the Pentagon widened its blacklist to include 188 firms, illustrating a growing concern about Chinese military connections across various sectors, including AI. The increasing number of entities on this list signals a heightened vigilance on the part of the US government, yet the enforcement of these controls continues to falter due to the software’s intangible nature. Companies like Anthropic are taking a firmer stance by outright banning access by Chinese firms and advocating for robust regulatory measures.

Such developments could provoke a reckoning in how the US manages its technological assets in a global context. With lawmakers voicing the need for software export controls analogous to semiconductor policies, the path towards more effective legislation appears fraught with complexity. The implications of these regulatory gaps are far-reaching; they not only affect national security but also shape the competitive landscape of AI technologies worldwide.

This material is informational and not a financial recommendation.