Algorithmic Cold War: China Fiercely Rejects US Accusations of Stealing AI Weights

Specific methods of training neural networks have become the cause of an international diplomatic scandal. On September 9, 2026, Beijing officially rejected accusations from the FBI, NSA, and CISA that Chinese corporations (DeepSeek, Alibaba, Moonshot AI) were engaged in large-scale "AI model distillation" of American frontier models.

The US claims that Asian startups illegally fed prompts into the APIs of Western systems to quickly and cheaply train their own algorithms. Chinese authorities called these statements baseless, emphasizing their own merits in R&D. In reality, this conflict means one thing: API outputs are now guarded by Washington as paranoiacally as blueprints for nuclear submarines. The technological wall between East and West is becoming absolutely impenetrable, and the terms of service for cloud AI platforms will be tightened at the government level to cut off "spy" requests.

Source: Chinese Government / US Intelligence / AP