Anthropic's CEO, Dario Amodei, explained why his company declined to sign a letter urging the White House not to ban Chinese open-weight AI models, even though more than 50 others are on board. In a blog post, he pointed out that while these models are cheaper, they are harder to control, which could lead to serious misuse by governments such as China. Amodei warned that open-weight models make it easier for bad actors to launch cyberattacks or even create biological threats. He suggested steps like limiting US chip exports to China and running strict safety tests on all models. His stance has split the industry: companies such as NVIDIA and Microsoft love open systems for their flexibility, but Anthropic is pushing for tighter rules to keep things safe.