# UbuntuGuard: A Culturally-Grounded Policy Benchmark for Equitable AI Safety in African Languages

arXiv: https://arxiv.org/abs/2601.12696 · HTML: https://mlsanity.com/paper/2601.12696/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- Builds on 16 of its 31 references that are in the map.
- First posted on arXiv 19 Jan 2026.

## Builds on (most cited first, 16 of 16)

- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [Ethical and social risks of harm from Language Models](https://mlsanity.com/paper/2112.04359/) (2021)
- [Improving alignment of dialogue agents via targeted human judgements](https://mlsanity.com/paper/2209.14375/) (2022)
- [WildGuard: Open One-Stop Moderation Tools for Safety Risks, Jailbreaks, and Refusals of LLMs](https://mlsanity.com/paper/2406.18495/) (2024)
- [NeMo Guardrails: A Toolkit for Controllable and Safe LLM Applications with Programmable Rails](https://mlsanity.com/paper/2310.10501/) (2023)
- [Low-Resource Languages Jailbreak GPT-4](https://mlsanity.com/paper/2310.02446/) (2023)
- [GEMBA-MQM: Detecting Translation Quality Error Spans with GPT-4](https://mlsanity.com/paper/2310.13988/) (2023)
- [Aegis2.0: A Diverse AI Safety Dataset and Risks Taxonomy for Alignment of LLM Guardrails](https://mlsanity.com/paper/2501.09004/) (2025)
- [PolyGuard: A Multilingual Safety Moderation Tool for 17 Languages](https://mlsanity.com/paper/2504.04377/) (2025)
- [CultureGuard: Towards Culturally-Aware Dataset and Guard Model for Multilingual Safety Applications](https://mlsanity.com/paper/2508.01710/) (2025)
- [DynaGuard: A Dynamic Guardian Model With User-Defined Policies](https://mlsanity.com/paper/2509.02563/) (2025)
- [Afrispeech-Dialog: A Benchmark Dataset for Spontaneous English Conversations in Healthcare and Beyond](https://mlsanity.com/paper/2502.03945/) (2025)
- [Beyond Contrastive Learning: Synthetic Data Enables List-wise Training with Multiple Levels of Relevance](https://mlsanity.com/paper/2503.23239/) (2025)
- [Amplify Initiative: Building A Localized Data Platform for Globalized AI](https://mlsanity.com/paper/2504.14105/) (2025)

## Built on by (most cited first, 5 of 5)

- [TukaBench: A Culturally Grounded Jailbreak Benchmark for African Languages](https://mlsanity.com/paper/2606.01322/) (2026)
- [The Illusion of Cross-Lingual Safety in Low-Resource Languages](https://mlsanity.com/paper/2608.11146/) (2026)
- [Language-Specific Gaps in AI Safety Training Datasets](https://mlsanity.com/paper/2608.13695/) (2026)
- [LLM Safety Alignment in Low-Resource Languages: A Systematic Literature Review](https://mlsanity.com/paper/2608.14626/) (2026)
- [A Translational Note on AI Safety Evaluation](https://mlsanity.com/paper/2609.06573/) (2026)

## Abstract (arXiv)

Current guardian models are predominantly Western-centric and optimized for high-resource languages, leaving low-resource African languages vulnerable to evolving harms, cross-lingual failures, and cultural misalignment. Moreover, most guardian models rely on rigid, predefined safety categories that fail to generalize across diverse linguistic and sociocultural contexts. Achieving robust safety requires flexible, runtime-enforceable policies and benchmarks that reflect local norms, harm scenarios, and cultural expectations. We introduce UbuntuGuard, the first policy-based safety benchmark for African languages built from adversarial queries authored by 155 domain experts across sensitive fields, including healthcare. From these expert-crafted queries, we derive context-specific safety policies and reference responses that capture culturally grounded risk signals, enabling policy-aligned evaluation of guardian models. We evaluate 15 models, comprising seven general-purpose LLMs and eight guardian models across three distinct variants: static, dynamic, and multilingual. Our findings reveal that existing English-centric benchmarks overestimate real-world multilingual safety, cross-lingual transfer provides partial but insufficient coverage, and dynamic models, while better equipped to leverage policies at inference time, still struggle to fully localize African-language contexts. These findings highlight the urgent need for multilingual, culturally grounded safety benchmarks to enable the development of reliable and equitable guardian models for low-resource languages.