COMPASS: A Framework for Evaluating Organization-Specific Policy Alignment in LLMs

• URL: http://arxiv.org/abs/2601.01836v1
• Date: Mon, 05 Jan 2026 06:57:45 GMT
• Title: COMPASS: A Framework for Evaluating Organization-Specific Policy Alignment in LLMs
• Abstract summary: We present the first systematic framework for evaluating whether large language models comply with organizational allowlist and denylist policies.<n>We show that models reliably handle legitimate requests but catastrophically fail at enforcing prohibitions, refusing only 13-40% of adversarial denylist violations.
• Score: 19.742967013586927
• License: http://creativecommons.org/licenses/by-nc-sa/4.0/
• Abstract: As large language models are deployed in high-stakes enterprise applications, from healthcare to finance, ensuring adherence to organization-specific policies has become essential. Yet existing safety evaluations focus exclusively on universal harms. We present COMPASS (Company/Organization Policy Alignment Assessment), the first systematic framework for evaluating whether LLMs comply with organizational allowlist and denylist policies. We apply COMPASS to eight diverse industry scenarios, generating and validating 5,920 queries that test both routine compliance and adversarial robustness through strategically designed edge cases. Evaluating seven state-of-the-art models, we uncover a fundamental asymmetry: models reliably handle legitimate requests (>95% accuracy) but catastrophically fail at enforcing prohibitions, refusing only 13-40% of adversarial denylist violations. These results demonstrate that current LLMs lack the robustness required for policy-critical deployments, establishing COMPASS as an essential evaluation framework for organizational AI safety.

Related papers

• A New Framework for Cybersecurity Refusals in AI Agents [52.94784168139071]
  We present the first framework for establishing refusal boundaries in offensive security contexts.<n>We apply this framework to assess how current LLM-powered agents adhere to appropriate refusal boundaries across a range of web-based offensive security scenarios.
  arXiv  Detail & Related papers  (2026-05-31T15:39:39Z)
• An Automated Framework for Cybersecurity Policy Compliance Assessment Against Security Control Standards [0.25489046505746704]
  PROPARAG is an audit support approach for evaluating organizational cybersecurity policies against security controls autonomously.<n>For each control, the approach retrieves relevant policy evidence, assesses coverage, identifies missing elements, and generates supporting explanations and recommendations.
  arXiv  Detail & Related papers  (2026-05-08T09:45:25Z)
• PolicyBank: Evolving Policy Understanding for LLM Agents [51.86716874651299]
  PolicyBank maintains structured, tool-level policy insights and iteratively refines them.<n>PolicyBank closes up to 82% of the gap toward a human oracle.
  arXiv  Detail & Related papers  (2026-04-16T20:29:30Z)
• Do LLMs Follow Their Own Rules? A Reflexive Audit of Self-Stated Safety Policies [0.0]
  LLMs internalize safety policies through RLHF, yet these policies are never formally specified and remain difficult to inspect.<n>Existing benchmarks evaluate models against external standards but do not measure whether models understand and enforce their own boundaries.<n>We introduce the Symbolic-Neural Consistency Audit (SNCA), a framework that extracts a model's self-stated safety rules via structured prompts.
  arXiv  Detail & Related papers  (2026-04-10T10:18:45Z)
• Security Assessment and Mitigation Strategies for Large Language Models: A Comprehensive Defensive Framework [0.6015898117103067]
  Large Language Models increasingly power critical infrastructure from healthcare to finance, yet their vulnerability to adversarial manipulation threatens system integrity and user safety.<n>This research addresses this gap by establishing a standardized vulnerability assessment framework and developing a multi-layered defensive system to protect against identified threats.
  arXiv  Detail & Related papers  (2026-03-17T20:32:06Z)
• OOD-MMSafe: Advancing MLLM Safety from Harmful Intent to Hidden Consequences [64.01706941950489]
  Current safety paradigms primarily target malicious intent or situational violations.<n>We propose shifting the safety frontier toward consequence-driven safety, a paradigm essential for the robust deployment of autonomous and embodied agents.<n>We develop the Consequence-Aware Safety Policy Optimization (CASPO) framework, which integrates the model's intrinsic reasoning as a dynamic reference for token-level self-distillation rewards.
  arXiv  Detail & Related papers  (2026-03-10T14:16:43Z)
• Towards Comprehensive Stage-wise Benchmarking of Large Language Models in Fact-Checking [64.97768177044355]
  Large Language Models (LLMs) are increasingly deployed in real-world fact-checking systems.<n>We present FactArena, a fully automated arena-style evaluation framework.<n>Our analyses reveal significant discrepancies between static claim-verification accuracy and end-to-end fact-checking competence.
  arXiv  Detail & Related papers  (2026-01-06T02:51:56Z)
• Syntactic Framing Fragility: An Audit of Robustness in LLM Ethical Decisions [1.0671844383558033]
  Large language models (LLMs) are increasingly deployed in consequential decision-making settings.<n>We study whether LLMs maintain consistent ethical judgments across logically equivalent but syntactically different prompts.<n>We introduce Syntactic Framing Fragility (SFF), a robustness evaluation framework that isolates purely syntactic effects.
  arXiv  Detail & Related papers  (2025-12-27T18:09:34Z)
• Training-Free Policy Violation Detection via Activation-Space Whitening in LLMs [21.5603664964501]
  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.
  arXiv  Detail & Related papers  (2025-12-03T17:23:39Z)
• Pluralistic Behavior Suite: Stress-Testing Multi-Turn Adherence to Custom Behavioral Policies [18.428149174461264]
  We present PBSUITE, a dynamic evaluation suite designed to assess large language models' capacity to adhere to pluralistic alignment specifications.<n>We find that leading open- and closed-source LLMs maintain robust adherence to behavioral policies in single-turn settings, but their compliance weakens substantially in multi-turn adversarial interactions.
  arXiv  Detail & Related papers  (2025-11-07T06:43:01Z)
• HALF: Harm-Aware LLM Fairness Evaluation Aligned with Deployment [52.374772443536045]
  HALF (Harm-Aware LLM Fairness) is a framework that assesses model bias in realistic applications and weighs the outcomes by harm severity.<n>We show that HALF exposes a clear gap between previous benchmarking success and deployment readiness.
  arXiv  Detail & Related papers  (2025-10-14T07:13:26Z)
• Safety Compliance: Rethinking LLM Safety Reasoning through the Lens of Compliance [49.50518009960314]
  Existing safety methods rely on ad-hoc taxonomy and lack a rigorous, systematic protection.<n>We develop a new benchmark for safety compliance by generating realistic LLM safety scenarios seeded with legal statutes.<n>Our experiments demonstrate that the Compliance Reasoner achieves superior performance on the new benchmark.
  arXiv  Detail & Related papers  (2025-09-26T12:11:29Z)
• Evaluating LLM Agent Adherence to Hierarchical Safety Principles: A Lightweight Benchmark for Probing Foundational Controllability Components [0.0]
  This paper introduces a lightweight, interpretable benchmark to evaluate an agent's ability to uphold a high-level safety principle.<n>Our evaluation reveals two primary findings: (1) a quantifiable "cost of compliance" where safety constraints degrade task performance even when compliant solutions exist, and (2) an "illusion of compliance" where high adherence often masks task incompetence rather than choice.
  arXiv  Detail & Related papers  (2025-06-03T01:16:34Z)
• Context Reasoner: Incentivizing Reasoning Capability for Contextualized Privacy and Safety Compliance via Reinforcement Learning [53.92712851223158]
  We formulate safety and privacy issues into contextualized compliance problems following the Contextual Integrity (CI) theory.<n>Under the CI framework, we align our model with three critical regulatory standards: EU AI Act, and HIPAA.<n>We employ reinforcement learning (RL) with a rule-based reward to incentivize contextual reasoning capabilities while enhancing compliance with safety and privacy norms.
  arXiv  Detail & Related papers  (2025-05-20T16:40:09Z)
• SafeBench: A Safety Evaluation Framework for Multimodal Large Language Models [75.67623347512368]
  We propose toolns, a comprehensive framework designed for conducting safety evaluations of MLLMs. Our framework consists of a comprehensive harmful query dataset and an automated evaluation protocol. Based on our framework, we conducted large-scale experiments on 15 widely-used open-source MLLMs and 6 commercial MLLMs.
  arXiv  Detail & Related papers  (2024-10-24T17:14:40Z)
• SORRY-Bench: Systematically Evaluating Large Language Model Safety Refusal [64.9938658716425]
  SORRY-Bench is a proposed benchmark for evaluating large language models' (LLMs) ability to recognize and reject unsafe user requests.<n>First, existing methods often use coarse-grained taxonomy of unsafe topics, and are over-representing some fine-grained topics.<n>Second, linguistic characteristics and formatting of prompts are often overlooked, like different languages, dialects, and more -- which are only implicitly considered in many evaluations.
  arXiv  Detail & Related papers  (2024-06-20T17:56:07Z)
• Trustworthy LLMs: a Survey and Guideline for Evaluating Large Language Models' Alignment [35.42539816648068]
  This paper presents a comprehensive survey of key dimensions that are crucial to consider when assessing large language models (LLMs) The survey covers seven major categories of LLM trustworthiness: reliability, safety, fairness, resistance to misuse, explainability and reasoning, adherence to social norms, and robustness. Results indicate that, in general, more aligned models tend to perform better in terms of overall trustworthiness.
  arXiv  Detail & Related papers  (2023-08-10T06:43:44Z)
• Towards a multi-stakeholder value-based assessment framework for algorithmic systems [76.79703106646967]
  We develop a value-based assessment framework that visualizes closeness and tensions between values. We give guidelines on how to operationalize them, while opening up the evaluation and deliberation process to a wide range of stakeholders.
  arXiv  Detail & Related papers  (2022-05-09T19:28:32Z)

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.