Cross-Lingual Bias in Large Language Models: A Comparative Analysis of English and Swahili

A new study from arXiv (2608.03532v1) finds that social biases in large language models transform rather than transfer across languages, with stereotype rates shifting by up to 12 percentage points on specific axes when comparing 4,900 English–Swahili prompt pairs submitted to GPT-5.2 and Gemini 2.5 Flash. Gemini's neutral-sentiment rate doubled in Swahili, GPT-5.2 refused 169 prompts in English and zero in Swahili, and over 55% of prompt pairs produced semantically dissimilar completions, indicating that English-only bias audits are inadequate for multilingual deployment.

arXiv:2608.03532v1 Announce Type: new Abstract: Large language models are increasingly deployed in multilingual contexts, yet safety alignment and bias evaluation remain overwhelmingly English-centric. We investigate whether social biases generalise across languages by submitting 4,900 symmetric English--Swahili prompt pairs to GPT-5.2 and Gemini 2.5 Flash across nine demographic bias axes, yielding 19,600 completions evaluated for stereotype prevalence, sentiment, refusal behaviour, and cross-lingual semantic similarity. Our findings show that bias transforms rather than transfers: stereotype rates shifted by up to 12 percentage points on specific axes, Gemini's neutral-sentiment rate doubled in Swahili, and GPT-5.2 refused 169 prompts in English and zero in Swahili, consistent with refusal behaviour anchored to English-language surface forms at the behavioural level. Over 55% of prompt pairs produced semantically dissimilar completions across both models. These reinforce the idea that English-only bias audits do not produce adequate coverage for multilingual deployment.