SHAPE: Unifying Safety, Helpfulness and Pedagogy for Educational LLMs

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Zhao, Sihang, Yu, Kangrui, Yuan, Youliang, He, Pinjia, Wen, Hongyi
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866918473840459776
author Zhao, Sihang
Yu, Kangrui
Yuan, Youliang
He, Pinjia
Wen, Hongyi
author_facet Zhao, Sihang
Yu, Kangrui
Yuan, Youliang
He, Pinjia
Wen, Hongyi
contents Large Language Models (LLMs) have been widely explored in educational scenarios. We identify a critical vulnerability in current educational LLMs, pedagogical jailbreaks, where students use answer-inducing prompts to elicit solutions rather than scaffolded instructions. To enable systematic study, we unify and formalize safe, helpful, and pedagogical behaviors with a knowledge-mastery graph and introduce SHAPE, a benchmark of 9,087 student-question pairs for evaluating tutoring behavior under adversarial pressure. We propose a graph-augmented tutoring pipeline that infers prerequisite concepts from queries, identifies mastery gaps, and routes generation between instructing and problem-solving via explicit gating. Experiments across multiple LLMs show that our method yields significantly improved safety under two pedagogical jailbreak settings, while maintaining near-ceiling helpfulness under the same evaluation protocol. Our code and data are available at https://github.com/MAPS-research/SHaPE
format Preprint
id arxiv_https___arxiv_org_abs_2604_22134
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle SHAPE: Unifying Safety, Helpfulness and Pedagogy for Educational LLMs
Zhao, Sihang
Yu, Kangrui
Yuan, Youliang
He, Pinjia
Wen, Hongyi
Computation and Language
Large Language Models (LLMs) have been widely explored in educational scenarios. We identify a critical vulnerability in current educational LLMs, pedagogical jailbreaks, where students use answer-inducing prompts to elicit solutions rather than scaffolded instructions. To enable systematic study, we unify and formalize safe, helpful, and pedagogical behaviors with a knowledge-mastery graph and introduce SHAPE, a benchmark of 9,087 student-question pairs for evaluating tutoring behavior under adversarial pressure. We propose a graph-augmented tutoring pipeline that infers prerequisite concepts from queries, identifies mastery gaps, and routes generation between instructing and problem-solving via explicit gating. Experiments across multiple LLMs show that our method yields significantly improved safety under two pedagogical jailbreak settings, while maintaining near-ceiling helpfulness under the same evaluation protocol. Our code and data are available at https://github.com/MAPS-research/SHaPE
title SHAPE: Unifying Safety, Helpfulness and Pedagogy for Educational LLMs
topic Computation and Language
url https://arxiv.org/abs/2604.22134