Bridging the Multilingual Safety Divide: Efficient, Culturally-Aware Alignment for Global South Languages

TL;DR

This paper highlights the challenges of ensuring safety and cultural appropriateness in multilingual large language models for Global South languages, emphasizing the need for localized, participatory safety strategies.

Contribution

It identifies key shortcomings of current safety approaches in low-resource and code-mixed languages and proposes a practical agenda for culturally-aware, participatory safety alignment.

Findings

Safety guardrails weaken on low-resource and code-mixed inputs

Culturally harmful behavior persists despite acceptable toxicity scores

English safety patches often do not transfer to low-resource languages

Abstract

Large language models (LLMs) are being deployed across the Global South, where everyday use involves low-resource languages, code-mixing, and culturally specific norms. Yet safety pipelines, benchmarks, and alignment still largely target English and a handful of high-resource languages, implicitly assuming safety and factuality ''transfer'' across languages. Evidence increasingly shows they do not. We synthesize recent findings indicating that (i) safety guardrails weaken sharply on low-resource and code-mixed inputs, (ii) culturally harmful behavior can persist even when standard toxicity scores look acceptable, and (iii) English-only knowledge edits and safety patches often fail to carry over to low-resource languages. In response, we outline a practical agenda for researchers and students in the Global South: parameter-efficient safety steering, culturally grounded evaluation and preference data, and participatory workflows that empower local communities to define and mitigate harm. Our aim is to make multilingual safety a core requirement-not an add-on-for equitable AI in underrepresented regions.