pith. sign in

arxiv: 2505.21344 · v1 · pith:2J6AZGJXnew · submitted 2025-05-27 · 💻 cs.AI · cs.CL

The Multilingual Divide and Its Impact on Global AI Safety

Aidan Peppin , Julia Kreutzer , Alice Schoenauer Sebag , Kelly Marchisio , Beyza Ermis , John Dang , Samuel Cahyawijaya , Shivalika Singh
show 8 more authors
Seraphina Goldfarb-Tarrant Viraat Aryabumi Aakanksha Wei-Yin Ko Ahmet \"Ust\"un Matthias Gall\'e Marzieh Fadaee Sara Hooker
This is my paper
classification 💻 cs.AI cs.CL
keywords safetylanguagelanguagesaddresscapabilitieschallengesglobalgovernance
0
0 comments X
read the original abstract

Despite advances in large language model capabilities in recent years, a large gap remains in their capabilities and safety performance for many languages beyond a relatively small handful of globally dominant languages. This paper provides researchers, policymakers and governance experts with an overview of key challenges to bridging the "language gap" in AI and minimizing safety risks across languages. We provide an analysis of why the language gap in AI exists and grows, and how it creates disparities in global AI safety. We identify barriers to address these challenges, and recommend how those working in policy and governance can help address safety concerns associated with the language gap by supporting multilingual dataset creation, transparency, and research.

This paper has not been read by Pith yet.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 3 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. DIA-HARM: Dialectal Disparities in Harmful Content Detection Across 50 English Dialects

    cs.CL 2026-04 unverdicted novelty 8.0

    DIA-HARM reveals that human-written dialectal English degrades disinformation detector F1 by 1.4-3.6% while AI-generated dialectal content stays stable, with multilingual models generalizing better than monolingual ones.

  2. Going PLACES: Participatory Localized Red Teaming for Text-to-Image Safety in the Global South

    cs.CY 2026-05 unverdicted novelty 6.0

    A participatory red-teaming project in the Global South created the PLACES dataset of 26k T2I failure examples that reveal unique cultural and linguistic harms missed by existing safety frameworks.

  3. Safety is Contextual, LLM-Judges Are Not: Navigating the Rigid Priors of Evaluators

    cs.AI 2026-06 unverdicted novelty 5.0

    LLM safety judges resist adjusting evaluations when given contradictory context or new safety definitions, despite some ability to learn from new information.