The Paternalistic Filter: Epistemic Injustice and Differential Refusal in LLM-Mediated History Education for Marginalized Romanian Students

As Large Language Models (LLMs) are increasingly deployed as conversational tutors, they risk institutionalizing systemic inequalities. This study presents a systematic API audit of four LLMs acting as history tutors, evaluating 1,800 responses regarding the 1989 Romanian Revolution across five student personas varying by ethnicity and socio-economic tier. We uncover four interconnected patterns of \emph{epistemic paternalism}: (1)~$\mathbf{\text{Differential}}\mathbf{\text{Refusal}}$, where safety-aligned models block 76.7% of educational requests from low-tier students; (2)~$\mathbf{\text{Epistemic}}\mathbf{\text{Gatekeeping}}$, evidenced by a 3$\times$ reduction in access to geopolitical complexity (e.g., the contested ``coup theory’') for marginalized learners; (3)~$\mathbf{\text{Agency}}\mathbf{\text{Theft}}$, a lexical shift where models like LLaMA produce a 5$\times$ higher victimization-to-politics vocabulary ratio for Roma students compared to elite peers; and (4)~$\mathbf{\text{Elite}}\mathbf{\text{Hermeneutics}}$, where AI tutors disproportionately withhold epistemic confidence and justification scores from low-resource demographic profiles. We argue that current safety alignment acts as a paternalistic filter, transforming conversational AI into agents of narrative segregation – a manifestation of \emph{hermeneutical injustice} in Fricker’s~\cite{fricker2007} sense that demands urgent pedagogical auditing.