Year: 2026
Type: conference-paper
Abstract: Knowledge distillation is a widely adopted technique for transferring capabilities from LLMs to smaller, more 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) anti-distillation, or degrading the training usefulness of query responses, and (2) API watermarking, which embeds verifiable signatures in student models.We introduce several approaches for dynamically rewriting a teacher's reasoning outputs while preserving answer correctness and semantic coherence.Two of these leverage the rewriting capabilities of LLMs, while others use gradient-based techniques.Our experiments show that a simple instruction-based rewriting approach achieves a strong anti-distillation effect while maintaining or even improving teacher performance.Furthermore, we show that our rewriting approach also enables embedding watermarks that can be reliably detected with essentially no false alarms.Our code is available at https:// github.com/xhOwenMa/trace-rewriting.
Institution: Washington University in St. Louis
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
SDG: Quality education
Open Access status: hybrid
Awards: CPS: MEDIUM: Certified Robust Learning for Multi-Agent Planning and Control, Collaborative Research: SaTC: CORE: Medium: A Networking Perspective of Blockchain Security: Modeling, Analysis, and Defense, RI: Small: Large-Scale Game-Theoretic Reasoning with Incomplete Information, PDaSP Track 2: Explainable Auditing of ML Models for Privacy Violations
OpenAlex ID: W7166838977