Reuters reported that two Chinese artificial‑intelligence firms, DeepSeek and Moonshot, built much of their new models on OpenAI’s GPT‑6 Astra output. The companies employed a technique called model distillation, copying knowledge from the larger system into smaller architectures. According to the report, the practice allowed them to accelerate development without training from scratch. American AI companies criticized the approach, describing it as a shortcut that merely reproduces advanced capabilities. The criticism suggests concerns about intellectual‑property and competitive fairness. The situation reflects growing tension between U.S. and Chinese players in the generative‑AI race.