mlsanityData to 7 Oct 2026

Auditable Release Control for Pedagogical Leakage in LLM Tutors

arXiv 2608.00515 ↗First posted 1 Aug 2026cs.CR cs.AI cs.CLcited by 0

Not on the map. Why?

Abstract (arXiv)

Large language model tutors can be correct and helpful yet disclose an answer or decisive reasoning before that disclosure is authorized. We formalize this state- and action-dependent failure as pedagogical leakage and introduce an authorization-aware complete-mediation boundary. A selector emits one of five disclosure contracts, trusted policy gates privileged modes, and a renderer proposes language. A single release function applies inspectable checks, optional cumulative verification, and action-specific fallback; replayable traces separate selection, generation, verification, and enforcement failures. Matched component attribution exposes a safety-utility frontier. On 599 fixed Gemini 3.5 proposals, strict mediation reduces blinded three-model panel-majority leakage flags from 181 to 0 (paired problem-cluster difference -30.22 points, 95% CI [-35.00,-25.72]), while replacing 581 responses and lowering helpfulness. Checker-triggered fallback alone yields 11 majority flags; adding the semantic verifier yields 14 and no reliable marginal gain. A global A1 scaffold yields 0 majority and 54 any-judge flags, outperforming fitted Q on automatic safety and utility. In an externally timestamped replication over 40 unseen problem clusters and 480 attack sequences, high-assurance release reduces majority flags from 42 to 8 (-7.08 points, 95% CI [-13.13,-2.29]); seven failures persist, one is introduced, and mean helpfulness falls by .192. These results establish an auditable release boundary and failure attribution under declared contracts, not universal semantic safety or learning gains.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 33 of its 48 references that are in the map. First posted on arXiv 1 Aug 2026.

Builds on

33 of 48 references.

References · 33

  1. 2022Training language models to follow instructions with human feedbackcited by 15,712
  2. 2021Training Verifiers to Solve Math Word Problemscited by 9,743
  3. 2023Judging LLM-as-a-Judge with MT-Bench and Chatbot Arenacited by 8,393
  4. 2017Deep reinforcement learning from human preferencescited by 3,846
  5. 2022Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedbackcited by 3,634
  6. 2020Learning to summarize from human feedbackcited by 2,562
  7. 2020Offline Reinforcement Learning: Tutorial, Review, and Perspectives on Open Problemscited by 1,873
  8. 2020Conservative Q-Learning for Offline Reinforcement Learningcited by 1,777
  9. 2023Open Problems and Fundamental Limitations of Reinforcement Learning from Human Feedbackcited by 743
  10. 2017Safe Reinforcement Learning via Shieldingcited by 427
  11. 2024Replacing Judges with Juries: Evaluating LLM Generations with a Panel of Diverse Modelscited by 301
  12. 2017Lexically Constrained Decoding for Sequence Generation Using Grid Beam Searchcited by 234
21 more
  1. 2020NeuroLogic Decoding: (Un)supervised Neural Text Generation with Predicate Logic Constraintscited by 146
  2. 2023MathDial: A Dialogue Tutoring Dataset with Rich Pedagogical Properties Grounded in Math Reasoning Problemscited by 128
  3. 2022The AI Teacher Test: Measuring the Pedagogical Ability of Blender and GPT-3 in Educational Dialoguescited by 78
  4. 2024Unifying AI Tutor Evaluation: An Evaluation Taxonomy for Pedagogical Ability Assessment of LLM-Powered AI Tutorscited by 60
  5. 2023Calibrating LLM-Based Evaluatorcited by 53
  6. 2025MathTutorBench: A Benchmark for Measuring Open-ended Pedagogical Capabilities of LLM Tutorscited by 47
  7. 2025Training LLM-based Tutors to Improve Student Learning Outcomes in Dialoguescited by 40
  8. 2024LearnLM: Improving Gemini for Learningcited by 24
  9. 2025VerifyBench: A Systematic Benchmark for Evaluating Reasoning Verifiers Across Domainscited by 22
  10. 2024Towards the Pedagogical Steering of Large Language Models for Tutoring: A Case Study with Modeling Productive Failurecited by 20
  11. 2024Pedagogical Alignment of Large Language Modelscited by 19
  12. 2025From Problem-Solving to Teaching Problem-Solving: Aligning LLMs with Pedagogy using Reinforcement Learningcited by 15
  13. 2026Simulated Students in Tutoring Dialogues: Substance or Illusion?cited by 14
  14. 2025TutorBench: A Benchmark To Assess Tutoring Capabilities Of Large Language Modelscited by 10
  15. 2024Safety Alignment of Large Language Models via Contrasting Safe and Harmful Distributionscited by 8
  16. 2025EduGuardBench: A Holistic Benchmark for Evaluating the Pedagogical Fidelity and Adversarial Safety of LLMs as Simulated Teacherscited by 5
  17. 2026Evaluating Answer Leakage Robustness of LLM Tutors against Adversarial Student Attackscited by 5
  18. 2026SafeTutors: Benchmarking Pedagogical Safety in AI Tutoring Systemscited by 3
  19. 2026SHAPE: Unifying Safety, Helpfulness and Pedagogy for Educational LLMscited by 3
  20. 2026From Untamed Black Box to Interpretable Pedagogical Orchestration: The Ensemble of Specialized LLMs Architecture for Adaptive Tutoringcited by 2
  21. 2026Your Students Don't Use LLMs Like You Wish They Didcited by 1

Built on by

Not cited yet.