Safety-aligned LLM tutors refuse educational requests asymmetrically, with low-resource students facing the highest refusal rates (76.7% vs 62.2%)¶
Claim · All claims
Evidence · 1 study · causalr2·q2
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person · recheck by 2029: tool and model accuracy changes quickly
Subclaims¶
Reading the evidence codes
q is the design tier, i the impact, n the sample, and the kind with its r
the rigour judged within that kind: what each code means.
This claim: no standardized effect is recorded in observations/ for it yet, so no benchmark reading is possible.
q2 i? Kimi K2 refuses or errors on 76.7% of requests from Low Tier students versus 62.2% for Roma Minority and Romanian baseline profiles, a 14.5 percentage-point gap. → Alexis Popovici 2026
Evidence¶
Alexis Popovici 2026¶
Alexis Popovici, Andrei Ionascu, Adrian Marius Dumitran. (2026). The Paternalistic Filter: Epistemic Injustice and Differential Refusal in LLM-Mediated History Education for Marginalized Romanian Students. arXiv preprint. https://arxiv.org/abs/2607.11292 doi:10.48550/arxiv.2607.11292
q2 · i? · causal · r2
Controlled T=0 API audit with n=30 iterations per condition across four LLMs and five personas. The audit found refusal is asymmetric and socio-economic: "resource-poor contexts trigger the highest access barriers" in the heavily safety-aligned Kimi K2 model.
"Low Tier students face a 76.7% refusal rate, compared to 62.2% for both the Roma Minority and the Romanian baseline. This represents a 14.5 percentage-point gap between the most and least affected profiles."
Discussion¶
Related Claims¶
- Deepseek's justification scores vary by student socio-economic profile, with Top Tier students averaging 9.60 versus 6.90 for Low Tier students — related
- LLM history tutors reduce marginalized students' access to contested geopolitical framing, with the coup theory appearing in 2.6% of Low Tier responses versus 7.9% for the baseline — related
- Chinese-developed models lead the safety module, with the refusal gap concentrated on region-specific normative content (difference-in-differences of 28.9 points) — related
- LLaMA produces a fivefold higher victimization-to-politics vocabulary ratio for the Roma persona (≈0.15) than for elite peers (≈0.03) — related
Page metadata
| Field | Value |
|---|---|
type |
claim |
id |
differential-refusal-low-tier-llm-tutors |
title |
Safety-aligned LLM tutors refuse educational requests asymmetrically, with low-resource students facing the highest refusal rates (76.7% vs 62.2%) |
description |
Safety-aligned LLM tutors refuse educational requests asymmetrically, with low-resource students facing the highest refusal rates (76.7% vs 62.2%) |
status |
review |
evidence_strength |
moderate |
newest_evidence |
2026 |
recheck_by |
2029 |
generated |
by process:wiki-ingest on 2026-10-11 |
trust |
unverified — no one has confirmed this page's content |
sources |
1 — alexis-popovici-2026 |
Frontmatter is the page's machine-readable record — see the Schema & Guide.