Why is the ethical line specifically on model distillation for you?

Was it ethical for Anthropic/OpenAI to train their models by gobbling a treasure trove of copyrighted material?