Skip to content
arXiv· arXiv:2604.03330· Mohammad Wali Ur Rahman, Martin Manuel Lopez, Lamia Tasnim Mim, Carter Farthing, Julius Battle, Kathryn Buckley, Salim Hariri·Indexed by this site · Originally published

AICCE: An AI-driven compliance checking engine

AICCE: AI Driven Compliance Checker Engine

AI summary

AICCE is a generative system combining RAG with dual-architecture reasoning for automated IPv6 compliance verification. The system encodes protocol standard semantics as vectors to retrieve specifications, and provides an interpretability mode with parallel LLM agent debate and a script execution mode that converts clauses into Python rules. Tested on IPv6 packet samples across 16 frontier generative models, accuracy and F1 reach up to 99%, and it can identify both regular and covert violations.

Read originalIEEE Transactions on Artificial Intelligence 2026