Knowledge Graph Representations for LLM-Based Policy Compliance Reasoning
For AI developers and regulators, this work provides a method to enhance LLM reasoning about AI policy compliance, though it is an incremental application of existing KG techniques to a new domain.
The paper presents an agentic framework that builds knowledge graphs from AI policy documents to improve LLM-based policy compliance reasoning, achieving consistent score improvements across five LLMs on 42 QA tasks spanning six reasoning types.
The risks posed by AI features are increasing as they are rapidly integrated into software applications. In response, regulations and standards for safe and secure AI have been proposed. In this paper, we present an agentic framework that constructs knowledge graphs (KGs) from AI policy documents and retrieves policy-relevant information to answer questions. We build KGs from three AI risk-related polices under two ontology schemas, and then evaluate five LLMs on 42 policy QA tasks spanning six reasoning types, from entity lookup to cross-policy inference, using both heuristic scoring and an LLM-as-judge. KG augmentation improves scores for all five models, and an open, LLM-discovered schema matches or exceeds the formal ontology.