Model distillation is an acceptable practice for training AI models

From the articleAnthropic says it cut Claude reasoning token theft by up to 99% in fight against industrial-scale distillationAlso in Anthropic says new safeguards sharply reduce AI distillation attacks | Reuters