Protecting Language Models Against Unauthorized Distillation through Trace Rewriting

ArXi:2602.15143v2 Announce Type: replace Knowledge distillation is a widely adopted technique for transferring capabilities from LLMs to smaller, efficient student models. However, unauthorized use of knowledge distillation takes unfair advantage of the considerable effort and cost put into developing frontier models. We investigate methods for modifying teacher-generated reasoning traces to achieve two objectives that deter unauthorized distillation: (1) \emph{anti-distillation}, or degrading the