How to Use AI to Help Find Civilian Harm

Between February twenty twenty-two and September twenty twenty-five, Bellingcat, the open-source investigative news organization, collected and documented more than twenty-five hundred incidents of civilian harm in Ukraine. To help identify such incidents faster, Bellingcat tested a machine learning model designed to scan Telegram posts and rank them by their likelihood of describing civilian casualties. The tool aims to streamline how investigators sift through massive amounts of conflict-related social media to spot potential evidence of war crimes and human rights violations.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton