I always feel that "distillation attack" is a forcibly created term. Distillation and attack sound very discordant. To rationalize this term, Anthropic divides AI labs into local and foreign, emphasizing that the victims and the justifiers are the "American" models. Such an announcement is clearly not written for the public; it reads more like an editorial, fulfilling a political task. But it is certain that users of Anthropic are almost exposed.