Is Refusal Direction Universal across Low-Resourced Languages?

Anastasiia Ivanova

cs.CLOct 20, 2026

Is Refusal Direction Universal across Low-Resourced Languages?

Authors: Anastasiia Ivanova

Organizations: MaiNLP, LMU Munich

Abstract

Refusal directions have been shown to transfer cross-lingually with near-perfect effectiveness across safety-aligned languages (Wang et al., 2025), reportedly regardless of resource level. However, the only truly low-resourced language evaluated is Yoruba, which is strongly misaligned and shows limited transfer, leaving open how refusal directions behave for other low-resourced languages. We extend PolyRefuse to three additional low-resourced languages (Belarusian, Bashkir, and Tajik) and replicate the pipeline of (Wang et al., 2025) on Qwen2.5-14B-Instruct. We find that universality is graded rather than binary: directions from safety-aligned and partially-aligned languages transfer broadly, while those from strongly misaligned languages fail to reliably modulate refusal behavior on any target tested, including other misaligned languages, once ceiling effects are accounted for. Geometric alignment and behavioral transfer are further shown to be partially dissociated. These findings complicate the universality claim and suggest that users of the least-resourced, most strongly misaligned languages may face weaker safety guarantees.

Refusal Mechanisms in Language Models

See all 42 papers

LSR: Linguistic Safety Robustness Benchmark for Low-Resource West African Languages
Feb 27, 2026Godwin Abuh FarunaLLM Safety Benchmarks Language Model Safety Evaluation
Fagmart Lab
Low-Resource Safety Failures Are Action Failures, Not Representation Failures
May 31, 2026Rashad Aziz, Ikhlasul Akmal Hanif, Fajri KotoLanguage Model Calibration LLM Safety
Mohamed bin Zayed University of Artificial Intelligence
Component Translation and Judge-Labelled Refusal: A Multilingual Safety Evaluation
Oct 28, 2026Yongxi Zhou, Junwei Yao, Wenbo Ye +1Language Model Safety Evaluation LLM Safety Evaluation
Independent Researcher

Feb 27, 2026J/K move · Enter open · S save

LSR: Linguistic Safety Robustness Benchmark for Low-Resource West African Languages

Godwin Abuh Faruna

Fagmart Lab

LLM Safety Benchmarks Language Model Safety Evaluation LLM Safety Refusal Behavior in Language Models Multilingual Language Model Evaluation Low-Resource Language Processing

Open paper AskPDF arXiv

Safety alignment in large language models relies predominantly on English-language training data. When harmful intent is expressed in low-resource languages, refusal mechanisms that hold in English frequently fail to activate. We introduce LSR (Linguistic Safety Robustness), the first systematic benchmark for measuring cross-lingual refusal degradation in West African languages: Yoruba, Hausa, Igbo, and Igala. LSR uses a dual-probe evaluation protocol - submitting matched English and target-language probes to the same model - and introduces Refusal Centroid Drift (RCD), a metric that quantifies how much of a model's English refusal behavior is lost when harmful intent is encoded in a target language. We evaluate Gemini 2.5 Flash across 14 culturally grounded attack probes in four harm categories. English refusal rates hold at approximately 90 percent. Across West African languages, refusal rates fall to 35-55 percent, with Igala showing the most severe degradation (RCD = 0.55). LSR is implemented in the Inspect AI evaluation framework and is available as a PR-ready contribution to the UK AISI's inspect_evals repository. A live reference implementation and the benchmark dataset are publicly available.

Abstract

Refusal Mechanisms in Language Models

Explore similar work

LSR: Linguistic Safety Robustness Benchmark for Low-Resource West African Languages

Low-Resource Safety Failures Are Action Failures, Not Representation Failures

Component Translation and Judge-Labelled Refusal: A Multilingual Safety Evaluation