“Respect my hair, white people!"

algorithmic racism in digital image databases

Abstract

This study aims to highlight how algorithmic racism manifests itself in digital image banks, particularly in representations of female beauty. It is a qualitative-quantitative and exploratory research, grounded in the concept of algorithmic racism and its relationship with structural racism, artificial intelligence, and data coloniality. For the analysis, the three main stock image websites used in Brazil were selected—Shutterstock, Getty Images, and Canva—based on searches for the keywords “beauty,” “skin care,” and “hair beauty.” From this dataset, content analysis was applied to the first results displayed by the platforms themselves, considering that such selections are determined by algorithms. The analyses show that most of the collected images depict white women with straight hair, contrary to the mixed-race reality and Black majority of the Brazilian population. These findings underscore the need to rethink how algorithms operate and their social—and especially racial—implications. Furthermore, it is essential to broaden the debate on media representations that perpetuate erasures and symbolic violence, as demonstrated here through algorithmic racism.