Chinese researchers and companies increasingly use outputs from leading American AI models to train and develop cheaper domestic alternatives, according to a Reuters investigation. The practice, often described as “distillation,” helps groups build comparable AI capabilities without needing direct access to the most advanced systems.

Outlets frame the implications in different ways. Digital Trends highlights concerns about intellectual property and the effectiveness of US technology restrictions. Rappler emphasizes how distillation can blunt the impact of chip curbs, suggesting China can still advance AI performance by leveraging US model outputs. Seeking Alpha centers on the same Reuters finding that export restrictions tightening coincide with greater reliance on American AI models for training.

Overall, the reporting focuses on the mechanism of using US model outputs to support Chinese development and on the broader policy and competitive questions it raises. The shared emphasis is that restrictions on chips and access may not fully prevent AI progress when training methods can rely on publicly obtainable or generated model outputs.