A new policy proposal suggests legalizing model distillation and scraping to help US AI models compete. Stratechery analyst Ben Thompson argues that the US should explicitly codify training data collection as fair use and ban anti-distillation terms of service. This discussion follows Alibaba's release of the open-weights Qwen 3.8 Max, a massive 2.4-trillion parameter model influenced by recent Chinese government calls for open-source collaboration.
Opening Kapyn…