mlsanityData to 14 Oct 2026

Safer Policy Compliance with Dynamic Epistemic Fallback

arXiv 2601.23094 ↗First posted 30 Jan 2026cs.CL cs.LGcited by 1

Not on the map. Why?

Abstract (arXiv)

Humans develop a series of cognitive defenses, known as epistemic vigilance, to combat risks of deception and misinformation from everyday interactions. Developing safeguards for LLMs inspired by this mechanism might be particularly helpful for their application in high-stakes tasks such as automating compliance with data privacy laws. In this paper, we introduce Dynamic Epistemic Fallback (DEF), a dynamic safety protocol for improving an LLM's inference-time defenses against deceptive attacks that make use of maliciously perturbed policy texts. Through various levels of one-sentence textual cues, DEF nudges LLMs to flag inconsistencies, refuse compliance, and fallback to their parametric knowledge upon encountering perturbed policy texts. Using globally recognized legal policies such as HIPAA and GDPR, our empirical evaluations report that DEF effectively improves the capability of frontier LLMs to detect and refuse perturbed versions of policies, with DeepSeek-R1 achieving a 100% detection rate in one setting. This work encourages further efforts to develop cognitively inspired defenses to improve LLM robustness against forms of harm and deception that exploit legal artifacts.

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Builds on 36 of its 47 references that are in the map. First posted on arXiv 30 Jan 2026.

Builds on

36 of 47 references.

References · 36

  1. 2022Training language models to follow instructions with human feedbackcited by 15,971
  2. 2025Qwen3 Technical Reportcited by 8,820
  3. 2023Universal and Transferable Adversarial Attacks on Aligned Language Modelscited by 3,104
  4. 2022Constitutional AI: Harmlessness from AI Feedbackcited by 3,050
  5. 2021TruthfulQA: Measuring How Models Mimic Human Falsehoodscited by 2,975
  6. 2016Concrete Problems in AI Safetycited by 1,642
  7. 2023Not what you've signed up for: Compromising Real-World LLM-Integrated Applications with Indirect Prompt Injectioncited by 1,443
  8. 2022Red Teaming Language Models with Language Modelscited by 1,026
  9. 2023Safe RLHF: Safe Reinforcement Learning from Human Feedbackcited by 656
  10. 2024Jailbreaking Leading Safety-Aligned LLMs with Simple Adaptive Attackscited by 486
  11. 2023Safety-Tuned LLaMAs: Lessons From Improving the Safety of Large Language Models that Follow Instructionscited by 387
  12. 2025Monitoring Reasoning Models for Misbehavior and the Risks of Promoting Obfuscationcited by 384
24 more
  1. 2023Adaptive Chameleon or Stubborn Sloth: Revealing the Behavior of Large Language Models in Knowledge Conflictscited by 341
  2. 2024Deliberative Alignment: Reasoning Enables Safer Language Modelscited by 308
  3. 2021Unsolved Problems in ML Safetycited by 292
  4. 2025Chain of Thought Monitorability: A New and Fragile Opportunity for AI Safetycited by 249
  5. 2023Poisoning Retrieval Corpora by Injecting Adversarial Passagescited by 159
  6. 2023Shifting Attention to Relevance: Towards the Predictive Uncertainty Quantification of Free-Form Large Language Modelscited by 150
  7. 2022Rich Knowledge Sources Bring Complex Knowledge Conflicts: Recalibrating Models to Reflect Conflicting Evidencecited by 144
  8. 2024To Believe or Not to Believe Your LLMcited by 120
  9. 2023Decomposing Uncertainty for Large Language Models through Input Clarification Ensemblingcited by 114
  10. 2023Navigating the Grey Area: How Expressions of Uncertainty and Overconfidence Affect Language Modelscited by 91
  11. 2025International AI Safety Reportcited by 67
  12. 2025Monitoring Monitorabilitycited by 58
  13. 2021Attacking Open-domain Question Answering by Injecting Misinformationcited by 41
  14. 2023Look Before You Leap: An Exploratory Study of Uncertainty Measurement for Large Language Modelscited by 41
  15. 2024GoldCoin: Grounding Large Language Models in Privacy Laws via Contextual Integrity Theorycited by 37
  16. 2025PrivaCI-Bench: Evaluating Privacy with Contextual Integrity and Legal Compliancecited by 37
  17. 2025Revisiting Epistemic Markers in Confidence Estimation: Can Markers Accurately Reflect Large Language Models' Uncertainty?cited by 16
  18. 2024From Loops to Oops: Fallback Behaviors of Language Models Under Uncertaintycited by 14
  19. 2024Epistemic Integrity in Large Language Modelscited by 7
  20. 2025Standardizing Intelligence: Aligning Generative AI for Regulatory and Operational Compliancecited by 5
  21. 2025On Epistemic Uncertainty of Visual Tokens for Object Hallucinations in Large Vision-Language Modelscited by 5
  22. 2025Scaling Policy Compliance Assessment in Language Models with Policy Reasoning Tracescited by 4
  23. 2024The Decoy Dilemma in Online Medical Information Evaluation: A Comparative Study of Credibility Assessments by LLM and Human Judgescited by 2
  24. 2025Safety Compliance: Rethinking LLM Safety Reasoning through the Lens of Compliancecited by 2

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026Sycophancy Undermines Epistemic Vigilance in Cooperative Vision-Language Taskscited by 2

Similar, not cited

Close in wording, no citation either way

  1. 2025Context Reasoner: Incentivizing Reasoning Capability for Contextualized Privacy and Safety Compliance via Reinforcement Learning
  2. 2026Safer Reasoning Traces: Measuring and Mitigating Chain-of-Thought Leakage in LLMs
  3. 2025Contextual Integrity in LLMs via Reasoning and Reinforcement Learning
  4. 2026CourtGuard: A Model-Agnostic Framework for Zero-Shot Policy Adaptation in LLM Safety
  5. 2026SafeThinker: Reasoning about Risk to Deepen Safety Beyond Shallow Alignment
  6. 2026Do LLMs Follow Their Own Rules? A Reflexive Audit of Self-Stated Safety Policies
  7. 2026Chain of Risk: Safety Failures in Large Reasoning Models and Mitigation via Adaptive Multi-Principle Steering
  8. 2025Trustworthy Reasoning: Evaluating and Enhancing Factual Accuracy in LLM Intermediate Thought Processes