PolicyMem: Geometric Policy Memory for LLM Governance

arXiv:2609.13734v1 Announce Type: cross Abstract: As large language models (LLMs) are increasingly deployed in real-world high-stakes applications, effective governance has become essential. Existing safeguards largely follow two paradigms: learning-based guards provide strong semantic discrimination but couple policy behavior to trained models and taxonomies, while programmable frameworks offer flexible control but require substantial manual prompt and workflow engineering. Neither externalizes policies as reusable operational states, making it difficult to consistently reuse policy evidence across detection, intervention, and verification. In this paper, we introduce PolicyMem, a geometric policy memory that externalizes natural-language policies as reusable geometric memory objects represented by low-rank subspaces in a shared representation space. A memory writer compiles natural-language policies into policy memory slots, and query-response pairs read the policy memory through projection energy. The resulting policy-evidence profile directly mediates the safety verdict and is reused for policy attribution and post-intervention verification. Coupled with a response rewriter, PolicyMem enables a detect-rewrite-verify loop for LLM governance. Across five widely used benchmarks, PolicyMem achieves state-of-the-art unsafe behavior detection while enabling effective policy attribution, rewriting, and post-intervention verification through the shared policy memory.

Related

• Policy-Grounded Safety Evaluation of 20 Large Language Models
• SpecAlign: Efficient Specification-Grounded Alignment of Large Language Models via Synthetic Data
• The Governance of Human-LLM Interaction: Safety Gating, Civility Steering, and Affective Default Lock-In
• An Information-Geometric Framework for Stability Analysis of Large Language Models under Entropic Stress

Source: arXiv cs.AI | 2026-09-15