New Report: 2026 Guide to Community-Led AI Audits
Our new Guide to Community-Led AI Audits documents six independent audits conducted across four countries — audits that uncovered a parole algorithm producing statistically random outcomes for 7,713 prisoners; a facial recognition system that classified adult women with Down Syndrome as children as young as five; and a ride-hailing platform that was simply unavailable 27% of the time in Roma neighborhoods, while always available everywhere else.
These are not edge cases. They are what AI governance looks like when communities are left out of the equation.
This guide shows what happens when they aren't — and gives advocates, funders, policymakers, and civil society organizations a tested, replicable model for meaningful AI accountability.
