SafeTutors: Benchmarking Pedagogical Safety in AI Tutoring Systems

• URL: http://arxiv.org/abs/2603.17373v1
• Date: Wed, 18 Mar 2026 05:33:50 GMT
• Title: SafeTutors: Benchmarking Pedagogical Safety in AI Tutoring Systems
• Abstract summary: We introduce SafeTutors, a benchmark that jointly evaluates safety and pedagogy across mathematics, physics, and chemistry.<n>We uncover that all models show broad harm; scale doesn't reliably help; and multi-turn dialogue worsens behavior, with pedagogical failures rising from 17.7% to 77.8%.<n>Harms also vary by subject, so mitigations must be discipline-aware, and single-turn "safe/helpful" results can mask systematic tutor failure over extended interaction.
• Score: 28.55608350602746
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Large language models are rapidly being deployed as AI tutors, yet current evaluation paradigms assess problem-solving accuracy and generic safety in isolation, failing to capture whether a model is simultaneously pedagogically effective and safe across student-tutor interaction. We argue that tutoring safety is fundamentally different from conventional LLM safety: the primary risk is not toxic content but the quiet erosion of learning through answer over-disclosure, misconception reinforcement, and the abdication of scaffolding. To systematically study this failure mode, we introduce SafeTutors, a benchmark that jointly evaluates safety and pedagogy across mathematics, physics, and chemistry. SafeTutors is organized around a theoretically grounded risk taxonomy comprising 11 harm dimensions and 48 sub-risks drawn from learning-science literature. We uncover that all models show broad harm; scale doesn't reliably help; and multi-turn dialogue worsens behavior, with pedagogical failures rising from 17.7% to 77.8%. Harms also vary by subject, so mitigations must be discipline-aware, and single-turn "safe/helpful" results can mask systematic tutor failure over extended interaction.

Related papers

• Dialectics of Alignment: Harnessing Unsafe Knowledge for Dynamic Safety Routing [27.31688887825119]
  This work challenges the orthodoxy that unsafe data must be discarded.<n>We introduce SafeMoE, a framework that isolates unsafe knowledge into domain-specific LowRank Adapters (LoRA experts) trained exclusively on harmful corpora.<n>Our findings suggest a paradigm in alignment: true safety requires not the masking of unsafe knowledge, but a shift of controlled integration.
  arXiv  Detail & Related papers  (2026-05-30T11:49:42Z)
• Pedagogical Safety in Educational Reinforcement Learning: Formalizing and Detecting Reward Hacking in AI Tutoring Systems [1.4323566945483497]
  Reinforcement learning (RL) is increasingly used to personalize instruction in intelligent tutoring systems.<n>We introduce a four-layer model of pedagogical safety for educational RL comprising structural, progress, behavioral, and alignment safety.<n>We evaluate the framework in a controlled simulation of an AI tutoring environment with 120 sessions across four conditions and three learner profiles, totaling 18,000 interactions.
  arXiv  Detail & Related papers  (2026-04-05T19:43:29Z)
• Robust Multimodal Safety via Conditional Decoding [52.92816441364308]
  Multimodal large-language models (MLLMs) often experience degraded safety alignment when harmful queries exploit cross-modal interactions.<n>We propose a simple conditional decoding strategy, CASA, that utilizes internal representations of MLLMs to predict a binary safety token before response generation.
  arXiv  Detail & Related papers  (2026-03-31T23:19:50Z)
• SafeSci: Safety Evaluation of Large Language Models in Science Domains and Beyond [134.43113804188195]
  We introduce SafeSci, a comprehensive framework for safety evaluation and enhancement in scientific contexts.<n>SafeSci comprises SafeSciBench, a multi-disciplinary benchmark with 0.25M samples, and SafeSciTrain, a large-scale dataset containing 1.5M samples for safety enhancement.
  arXiv  Detail & Related papers  (2026-03-02T08:16:04Z)
• CASTLE: A Comprehensive Benchmark for Evaluating Student-Tailored Personalized Safety in Large Language Models [55.0103764229311]
  We propose the concept of Student-Tailored Personalized Safety and construct CASTLE based on educational theories.<n>This benchmark covers 15 educational safety risks and 14 student attributes, comprising 92,908 bilingual scenarios.
  arXiv  Detail & Related papers  (2026-02-05T13:13:19Z)
• The Missing Half: Unveiling Training-time Implicit Safety Risks Beyond Deployment [148.80266237240713]
  implicit training-time safety risks are driven by a model's internal incentives and contextual background information.<n>We present the first systematic study of this problem, introducing a taxonomy with five risk levels, ten fine-grained risk categories, and three incentive types.<n>Our results identify an overlooked yet urgent safety challenge in training.
  arXiv  Detail & Related papers  (2026-02-04T04:23:58Z)
• EduGuardBench: A Holistic Benchmark for Evaluating the Pedagogical Fidelity and Adversarial Safety of LLMs as Simulated Teachers [8.123835490773095]
  Large Language Models for Simulating Professions (SP-LLMs) are pivotal for personalized education.<n>EduGuardBench assesses professional fidelity using a Role-playing Fidelity Score (RFS)<n>It also probes safety vulnerabilities using persona-based adversarial prompts targeting both general harms and, particularly, academic misconduct.
  arXiv  Detail & Related papers  (2025-11-10T09:42:24Z)
• SOSBENCH: Benchmarking Safety Alignment on Scientific Knowledge [11.63268709958876]
  SOSBench is a regulation-grounded, hazard-focused benchmark for large language models.<n>It covers six high-risk scientific domains: chemistry, biology, medicine, pharmacology, physics, and psychology.<n>We evaluate frontier models within a unified evaluation framework using our SOSBench.
  arXiv  Detail & Related papers  (2025-05-27T17:47:08Z)
• Safety Pretraining: Toward the Next Generation of Safe AI [68.99129474671282]
  We present a data-centric pretraining framework that builds safety into the model from the start.<n>Our framework consists of four key steps: Safety Filtering, Safety Rephrasing, Native Refusal and Harmfulness-Tag annotated pretraining.<n>Our safety-pretrained models reduce attack success rates from 38.8% to 8.4% on standard LLM safety benchmarks with no performance on general degradation tasks.
  arXiv  Detail & Related papers  (2025-04-23T17:58:08Z)
• Safe Vision-Language Models via Unsafe Weights Manipulation [75.04426753720551]
  We revise safety evaluation by introducing Safe-Ground, a new set of metrics that evaluate safety at different levels of granularity.<n>We take a different direction and explore whether it is possible to make a model safer without training, introducing Unsafe Weights Manipulation (UWM)<n>UWM uses a calibration set of safe and unsafe instances to compare activations between safe and unsafe content, identifying the most important parameters for processing the latter.
  arXiv  Detail & Related papers  (2025-03-14T17:00:22Z)
• ALERT: A Comprehensive Benchmark for Assessing Large Language Models' Safety through Red Teaming [64.86326523181553]
  ALERT is a large-scale benchmark to assess safety based on a novel fine-grained risk taxonomy. It aims to identify vulnerabilities, inform improvements, and enhance the overall safety of the language models.
  arXiv  Detail & Related papers  (2024-04-06T15:01:47Z)

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.