판 이력 — Hollow-LLM Attack: Computationally Trivial Weights in Zero-Knowledge Verification of LLM Inference | AIChainDay