According to CoinMeta, Zuckerberg published a lengthy article today in which he publicly defended model distillation. He believes that models learning from other models is inherently part of the open-source ecosystem. If the United States wants to continue to lead in open-source development, it must relax restrictions on distillation and training data. He also directly refuted the claim that "distillation is harmful": anything that can be observed should be allowed to be learned from. Zuckerberg pointed out that if the U.S. binds itself first, it will only allow other countries' open-source models to catch up faster. Contrary to the recent actions of OpenAI and Anthropic, these two companies are taking strict measures to prevent competitors from bypassing restrictions and large-scale capturing of model outputs. Zuckerberg believes that this should not lead to a restriction on learning between models as well. The latest Muse Glimmer released today was itself distilled using Muse Spark.