Training-Free Policy Violation Detection via Activation-Space Whitening in LLMs

• URL: http://arxiv.org/abs/2512.03994v1
• Date: Wed, 03 Dec 2025 17:23:39 GMT
• Title: Training-Free Policy Violation Detection via Activation-Space Whitening in LLMs
• Authors: Oren Rachmil, Roy Betser, Itay Gershon, Omer Hofman, Nitay Yakoby, Yuval Meron, Idan Yankelev, Asaf Shabtai, Yuval Elovici, Roman Vainshtein,
• Abstract summary: We propose a training-free and efficient method that treats policy violation detection as an out-of-distribution detection problem.<n>Inspired by whitening techniques, we apply a linear transformation to decorrelate the model's hidden activations and standardize them to zero mean and unit variance.<n>On a challenging policy benchmark, our approach achieves state-of-the-art results, surpassing both existing guardrails and fine-tuned reasoning models.
• Score: 21.5603664964501
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Aligning proprietary large language models (LLMs) with internal organizational policies has become an urgent priority as organizations increasingly deploy LLMs in sensitive domains such as legal support, finance, and medical services. Beyond generic safety filters, enterprises require reliable mechanisms to detect policy violations within their regulatory and operational frameworks, where breaches can trigger legal and reputational risks. Existing content moderation frameworks, such as guardrails, remain largely confined to the safety domain and lack the robustness to capture nuanced organizational policies. LLM-as-a-judge and fine-tuning approaches, though flexible, introduce significant latency and lack interpretability. To address these limitations, we propose a training-free and efficient method that treats policy violation detection as an out-of-distribution (OOD) detection problem. Inspired by whitening techniques, we apply a linear transformation to decorrelate the model's hidden activations and standardize them to zero mean and unit variance, yielding near-identity covariance matrix. In this transformed space, we use the Euclidean norm as a compliance score to detect policy violations. The method requires only the policy text and a small number of illustrative samples, which makes it light-weight and easily deployable. On a challenging policy benchmark, our approach achieves state-of-the-art results, surpassing both existing guardrails and fine-tuned reasoning models. This work provides organizations with a practical and statistically grounded framework for policy-aware oversight of LLMs, advancing the broader goal of deployable AI governance. Code is available at: https://tinyurl.com/policy-violation-detection

Related papers

• When to Act, Ask, or Learn: Uncertainty-Aware Policy Steering [10.01278648231868]
  Policy steering is an emerging way to adapt robot behaviors at deployment-time.<n> Vision-Language Models (VLMs) are promising general-purpose verifiers due to their reasoning capabilities.<n>We propose uncertainty-aware policy steering (UPS), a framework that jointly reasons about semantic task uncertainty and low-level action feasibility.
  arXiv  Detail & Related papers  (2026-02-25T23:23:22Z)
• GAVEL: Towards rule-based safety through activation monitoring [2.337566423505956]
  Large language models (LLMs) are increasingly paired with activation-based monitoring to detect and prevent harmful behaviors.<n>Existing activation safety approaches, trained on broad misuse datasets, struggle with poor precision, limited flexibility, and lack of interpretability.<n>This paper introduces a new paradigm: rule-based activation safety, inspired by rule-sharing practices in cybersecurity.
  arXiv  Detail & Related papers  (2026-01-27T16:31:39Z)
• Reasoning over Precedents Alongside Statutes: Case-Augmented Deliberative Alignment for LLM Safety [59.01189713115365]
  We evaluate the impact of explicitly specifying extensive safety codes versus demonstrating them through illustrative cases.<n>We find that referencing explicit codes inconsistently improves harmlessness and systematically degrades helpfulness.<n>We propose CADA, a case-augmented deliberative alignment method for LLMs utilizing reinforcement learning on self-generated safety reasoning chains.
  arXiv  Detail & Related papers  (2026-01-12T21:08:46Z)
• MAESTRO: Meta-learning Adaptive Estimation of Scalarization Trade-offs for Reward Optimization [56.074760766965085]
  Group-Relative Policy Optimization has emerged as an efficient paradigm for aligning Large Language Models (LLMs)<n>We propose MAESTRO, which treats reward scalarization as a dynamic latent policy, leveraging the model's terminal hidden states as a semantic bottleneck.<n>We formulate this as a contextual bandit problem within a bi-level optimization framework, where a lightweight Conductor network co-evolves with the policy by utilizing group-relative advantages as a meta-reward signal.
  arXiv  Detail & Related papers  (2026-01-12T05:02:48Z)
• Policy-Conditioned Policies for Multi-Agent Task Solving [53.67744322553693]
  In this work, we propose a paradigm shift that bridges the gap by representing policies as human-interpretable source code.<n>We reformulate the learning problem by utilizing Large Language Models (LLMs) as approximate interpreters.<n>We formalize this process as textitProgrammatic Iterated Best Response (PIBR), an algorithm where the policy code is optimized by textual gradients.
  arXiv  Detail & Related papers  (2025-12-24T07:42:10Z)
• MENTOR: A Metacognition-Driven Self-Evolution Framework for Uncovering and Mitigating Implicit Risks in LLMs on Domain Tasks [17.598413159363393]
  Current alignment efforts primarily target explicit risks such as bias, hate speech, and violence.<n>We propose MENTOR: A MEtacognition-driveN self-evoluTion framework for uncOvering and mitigating implicit risks in large language models.<n>We release a supporting dataset of 9,000 risk queries spanning education, finance, and management to enhance domain-specific risk identification.
  arXiv  Detail & Related papers  (2025-11-10T13:51:51Z)
• Safe and Compliant Cross-Market Trade Execution via Constrained RL and Zero-Knowledge Audits [0.5586191108738564]
  We present a cross-market algorithmic trading system that balances execution quality with rigorous compliance enforcement.<n>The architecture comprises a high-level planner, a reinforcement learning execution agent, and an independent compliance agent.<n>We report effects at the 95% confidence level using paired t-tests and examine tail risk via CVaR.
  arXiv  Detail & Related papers  (2025-10-06T15:52:12Z)
• Towards Policy-Compliant Agents: Learning Efficient Guardrails For Policy Violation Detection [25.53228630260007]
  PolicyGuardBench is a benchmark of about 60k examples for detecting policy violations in agent trajectories.<n>PolicyGuard-4B is a lightweight guardrail model that delivers strong detection accuracy across all tasks.<n>Together, PolicyGuardBench and PolicyGuard-4B provide the first comprehensive framework for studying policy compliance in web agent trajectories.
  arXiv  Detail & Related papers  (2025-10-03T20:03:19Z)
• Policy-as-Prompt: Turning AI Governance Rules into Guardrails for AI Agents [0.19336815376402716]
  We introduce a regulatory machine learning framework that converts unstructured design artifacts (like PRDs, TDDs, and code) into verifiable runtime guardrails.<n>Our Policy as Prompt method reads these documents and risk controls to build a source-linked policy tree.<n>System is built to enforce least privilege and data minimization.
  arXiv  Detail & Related papers  (2025-09-28T17:36:52Z)
• On the Design of KL-Regularized Policy Gradient Algorithms for LLM Reasoning [59.11784194183928]
  Policy gradient algorithms have been successfully applied to enhance the reasoning capabilities of large language models (LLMs)<n>Regularized Policy Gradient (RPG) view shows that the widely-used $k_3$ penalty is exactly the unnormalized KL.<n>RPG-REINFORCE with RPG-Style Clip improves accuracy by up to $+6$ absolute percentage points over DAPO.
  arXiv  Detail & Related papers  (2025-05-23T06:01:21Z)
• Conformal Policy Learning for Sensorimotor Control Under Distribution Shifts [61.929388479847525]
  This paper focuses on the problem of detecting and reacting to changes in the distribution of a sensorimotor controller's observables. The key idea is the design of switching policies that can take conformal quantiles as input. We show how to design such policies by using conformal quantiles to switch between base policies with different characteristics.
  arXiv  Detail & Related papers  (2023-11-02T17:59:30Z)
• Robust Multi-Agent Reinforcement Learning via Adversarial Regularization: Theoretical Foundation and Stable Algorithms [79.61176746380718]
  Multi-Agent Reinforcement Learning (MARL) has shown promising results across several domains. MARL policies often lack robustness and are sensitive to small changes in their environment. We show that we can gain robustness by controlling a policy's Lipschitz constant. We propose a new robust MARL framework, ERNIE, that promotes the Lipschitz continuity of the policies.
  arXiv  Detail & Related papers  (2023-10-16T20:14:06Z)
• Trust-Region-Free Policy Optimization for Stochastic Policies [60.52463923712565]
  We show that the trust region constraint over policies can be safely substituted by a trust-region-free constraint without compromising the underlying monotonic improvement guarantee. We call the resulting algorithm Trust-REgion-Free Policy Optimization (TREFree) explicit as it is free of any trust region constraints.
  arXiv  Detail & Related papers  (2023-02-15T23:10:06Z)
• Privacy-Constrained Policies via Mutual Information Regularized Policy Gradients [54.98496284653234]
  We consider the task of training a policy that maximizes reward while minimizing disclosure of certain sensitive state variables through the actions. We solve this problem by introducing a regularizer based on the mutual information between the sensitive state and the actions. We develop a model-based estimator for optimization of privacy-constrained policies.
  arXiv  Detail & Related papers  (2020-12-30T03:22:35Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.