Study: Hate speech on Chile's SOSAFE spikes in changing neighborhoods

A new study from Eduardo Graells-Garrido, Daniela Opitz, Francisco Rowe, and Carmen Cabrera examines how migration drives social exclusion at fine spatiotemporal scales. Analyzing over 550,000 geolocated reports from SOSAFE, Chile's largest citizen reporting platform, the researchers fine-tuned a Spanish hate speech classifier and validated it against human labels. Their key finding: reports mentioning migrants are significantly more likely to contain hate speech compared to other reports. However, that hate speech doesn't cluster in established migrant communities. Instead, it concentrates in neighborhoods experiencing recent demographic change, defined as post-2010 migrant arrivals, revealing a pattern of 'digital bordering' where exclusionary discourse forms around new integration pressures.

Spatial analysis confirmed that hate speech hotspots coincide with areas where recent migrants make up more than a third of the population, while coldspots appear in high-education sectors with minimal recent migration. The study also found that both hate speech reports and migrant-mentioning reports receive more engagement from other users, though combining the two doesn't amplify engagement further. This suggests the platform's mechanics may inadvertently reward divisive content. The findings matter for urban policy and platform design: citizen reporting tools meant to improve safety can become arenas where demographic change triggers hostility, with algorithmic feedback loops potentially deepening social fragmentation in vulnerable neighborhoods.