Bo Jiang
We have not lifted any functions out of this paper's repositories yet, so there is nothing we have run. If it links a repository, it is listed below.
Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.
Knowledge distillation from proprietary LLM APIs poses a growing threat to model providers, yet defenses against this attack remain fragmented and unevaluated. We present Distill-Guard, a framework for systematically evaluating output-level defenses against LLM knowledge distillation. We introduce a taxonomy of three defense categories-output perturbation, data poisoning, and information throttlingand evaluate nine defense configurations using a standardized pipeline with Qwen3-14B as teacher and Qwen2.5-7B-Instruct as student across three benchmarks (MATH-500, Hu-manEval+, MT-Bench). Our results reveal that, in a same-family distillation setting against a naive attacker, most output-level defenses are surprisingly ineffective: paraphrasing-based perturbation barely degrades distilled student quality, and data poisoning primarily impairs conversational fluency while leaving taskspecific capabilities intact. Only chain-ofthought removal substantially impairs mathematical reasoning (31.4% vs. 67.8% baseline), though code generation remains unaffected. These findings demonstrate that the effectiveness of distillation defenses is highly task-dependent and that current output-level approaches are insufficient to broadly prevent knowledge theft. * Corresponding author. Code and data will be released at https://github.com/bojiang/distillguard upon publication.
The same record, over MCP at https://syntology.ai/mcp:
get_harvested_code_for_paper("2603.07835")
get_code_for_paper("2603.07835")
have("2603.07835")
Connect an agent — have() is free.