Measuring Moral LLM Responses in Multilingual Capacities
This addresses the need to understand and guardrail LLM moral responses across languages, but it is incremental as it builds on existing benchmarking datasets.
The study evaluated frontier and open-source LLMs across five moral dimensions in low- and high-resource languages, finding GPT-5 performed best with averages like 4.73 in Harm Prevention, while Gemini 2.5 Pro scored as low as 1.39 in Consent & Autonomy.
With LLM usage becoming widespread across countries, languages, and humanity more broadly, the need to understand and guardrail their multilingual responses increases. Large-scale datasets for testing and benchmarking have been created to evaluate and facilitate LLM responses across multiple dimensions. In this study, we evaluate the responses of frontier and leading open-source models in five dimensions across low and high-resource languages to measure LLM accuracy and consistency across multilingual contexts. We evaluate the responses using a five-point grading rubric and a judge LLM. Our study shows that GPT-5 performed the best on average in each category, while other models displayed more inconsistency across language and category. Most notably, in the Consent & Autonomy and Harm Prevention & Safety categories, GPT scored the highest with averages of 3.56 and 4.73, while Gemini 2.5 Pro scored the lowest with averages of 1.39 and 1.98, respectively. These findings emphasize the need for further testing on how linguistic shifts impact LLM responses across various categories and improvement in these areas.