A position paper argues that current AI leaderboards are biased against regions in the Global South due to a lack of institutional design. The authors claim that high-quality benchmarks for languages such as Hindi and Swahili already exist but are not included on global leaderboards, which are often driven by commercial interests from the Global North. A consultation with 58 AI practitioners in India found a preference for formal governance and conflict management over relying solely on data collection. The paper suggests that regional leaderboards with independent governance could provide a more equitable solution. --- Why it matters: This matters to researchers because it highlights the need for more inclusive and representative AI evaluation metrics, which can impact the development of language models and other applications serving diverse populations. Source: https://arxiv.org/abs/2608.18117