mlsanityData to 14 Oct 2026

Privacy Reasoning in Ambiguous Contexts

arXiv 2506.12241 ↗First posted 13 Jun 2025cs.AI cs.LGcited by 11

On the map: LLM reasoning → Reasoning for agents, planning and code

also in 1 more
Abstract (arXiv)

We study the ability of language models to reason about appropriate information disclosure - a central aspect of the evolving field of agentic privacy. Whereas previous works have focused on evaluating a model's ability to align with human decisions, we examine the role of ambiguity and missing context on model performance when making information-sharing decisions. We identify context ambiguity as a crucial barrier for high performance in privacy assessments. By designing Camber, a framework for context disambiguation, we show that model-generated decision rationales can reveal ambiguities and that systematically disambiguating context based on these rationales leads to significant accuracy improvements (up to 13.3% in precision and up to 22.3% in recall) as well as reductions in prompt sensitivity. Overall, our results indicate that approaches for context disambiguation are a promising way forward to enhance agentic privacy reasoning.

Key facts. Cited by 11 papers in the map (data through 14 Oct 2026). Placed in Reasoning for agents, planning and code, one of 17 directions in LLM reasoning. Builds on 29 of its 35 references that are in the map. First posted on arXiv 13 Jun 2025.

Builds on

29 of 35 references.

Closest ancestors · 1

Same direction, newest first

  1. 2025Contextual Integrity in LLMs via Reasoning and Reinforcement Learning29 May

Other references · 28

  1. 2020Retrieval-Augmented Generation for Knowledge-Intensive NLP Taskscited by 9,297
  2. 2016Deep Learning with Differential Privacycited by 2,728
  3. 2020Extracting Training Data from Large Language Modelscited by 2,189
  4. 2021Did Aristotle Use a Laptop? A Question Answering Benchmark with Implicit Reasoning Strategiescited by 743
  5. 2023DecodingTrust: A Comprehensive Assessment of Trustworthiness in GPT Modelscited by 518
  6. 2025Reasoning Models Don't Always Say What They Thinkcited by 415
  7. 2021Large Language Models Can Be Strong Differentially Private Learnerscited by 412
  8. 2021Differentially Private Fine-tuning of Language Modelscited by 400
  9. 2024TrustLLM: Trustworthiness in Large Language Modelscited by 285
  10. 2023Can LLMs Keep a Secret? Testing Privacy Implications of Language Models via Contextual Integrity Theorycited by 230
  11. 2023MuSR: Testing the Limits of Chain-of-thought with Multistep Soft Reasoningcited by 201
  12. 2025ZebraLogic: On the Scaling Limits of LLMs for Logical Reasoningcited by 201
16 more
  1. 2023Beyond Memorization: Violating Privacy Via Inference with Large Language Modelscited by 175
  2. 2023FANToM: A Benchmark for Stress-testing Machine Theory of Mind in Interactionscited by 171
  3. 2024PrivacyLens: Evaluating Privacy Norm Awareness of Language Models in Actioncited by 171
  4. 2023Minding Language Models' (Lack of) Theory of Mind: A Plug-and-Play Multi-Character Belief Trackercited by 95
  5. 2024AirGapAgent: Protecting Privacy-Conscious Conversational Agentscited by 94
  6. 2025AgentDAM: Privacy Leakage Evaluation for Autonomous Web Agentscited by 82
  7. 2024A Chain-of-Thought Is as Strong as Its Weakest Link: A Benchmark for Verifiers of Reasoning Chainscited by 65
  8. 2025QuestBench: Can LLMs ask the right question to acquire information in reasoning tasks?cited by 63
  9. 2024MultiTrust: A Comprehensive Benchmark Towards Trustworthy Multimodal Large Language Modelscited by 53
  10. 2025On the Trustworthiness of Generative Foundation Models: Guideline, Assessment, and Perspectivecited by 41
  11. 2024GoldCoin: Grounding Large Language Models in Privacy Laws via Contextual Integrity Theorycited by 37
  12. 2024Operationalizing Contextual Integrity in Privacy-Conscious Assistantscited by 37
  13. 2024CI-Bench: Benchmarking Contextual Integrity of AI Assistants on Synthetic Datacited by 37
  14. 2025PrivaCI-Bench: Evaluating Privacy with Contextual Integrity and Legal Compliancecited by 37
  15. 2025Protecting Users From Themselves: Safeguarding Contextual Privacy in Interactions with Conversational Agentscited by 26
  16. 2024Privacy Checklist: Privacy Violation Detection Grounding on Contextual Integrity Theorycited by 18

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 3; 2026: 8.

11 papers.

  1. 2026Skill-Inject: Measuring Agent Vulnerability to Skill File Attackscited by 90
  2. 2026AI Agents May Always Fall for Prompt Injectionscited by 9
  3. 2025Beyond Data Privacy: New Privacy Risks for Large Language Modelscited by 6
  4. 2026LiSA: Lifelong Safety Adaptation via Conservative Policy Inductioncited by 3
  5. 2026Janus: a Playground for User-Involved Agentic Permission Managementcited by 3
  6. 2026PrivacyReasoner: Can LLM Emulate a Human-like Privacy Mind?cited by 2
  7. 2025Black Box Absorption: LLMs Undermining Innovative Ideascited by 1
  8. 2025MiniScope: A Least Privilege Framework for Authorizing Tool Calling Agentscited by 1
  9. 2026ContextLens: Modeling Imperfect Privacy and Safety Context for Legal Compliancecited by 0
  10. 2026Engineering Robustness into Personal Agents with the AI Workflow Storecited by 0
  11. 2026CIDER: A Dataset of Contextual Disclosure Boundaries for Privacy Preference Alignmentcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 20251-2-3 Check: Enhancing Contextual Privacy in LLM via Multi-Agent Reasoning
  2. 2026Loose Lips: Understanding Privacy Leakage in Agentic Dialogue
  3. 2026PrivacyAlign: Contextual Privacy Alignment for LLM Agents
  4. 2024Privacy Leakage Overshadowed by Views of AI: A Study on Human Oversight of Privacy in Language Model Agent
  5. 2025Beyond Verification: Abductive Explanations for Post-AI Assessment of Privacy Leakage
  6. 2025Personalizing Agent Privacy Decisions via Logical Entailment
  7. 2026MPCI-Bench: A Benchmark for Multimodal Pairwise Contextual Integrity Evaluation of Language Model Agents
  8. 2026AgentSCOPE: Evaluating Contextual Privacy Across Agentic Workflows