Tools

Everything here is built to be used, not admired.

These are the working parts of a single method.

The AI Lifecycle is the lens, where human rights enter each phase of building a system. The HRIA Workbook turns that lens into a document teams complete, stage by stage. Why human rights, not ethics is the conviction underneath it all. HumRights-Bench is the research edge, testing whether AI models can reason about rights at all. The African and Latin American toolboxes adapt the method with regional partners. And the Course teaches the whole approach from the ground up. Start anywhere, but they are stronger together.

The AI Lifecycle

Finding where rights enter the build.

The human rights-based approach mapped onto each phase of development, from objective-setting to deployment and audit.

HRIA Workbook

Working through human rights risk, stage by stage.

A six-stage Human Rights Impact Assessment that runs alongside the AI lifecycle, pairing documentation with reflection questions at every step.

Why Human Rights, Not Ethics

Choosing law you can act on.

A human rights-based approach gives teams a shared, agreed starting point that voluntary ethics principles cannot.

HumRights-Bench

Testing whether AI can reason about rights.

The first expert-validated benchmark for human rights reasoning in AI. Frontier models scored near chance.

African Toolbox

Adapting the method across Africa.

Built with African researchers: frugal, inventive, community-led AI the field should learn from, across health, agriculture, climate and language.

Latin American Toolbox

Adapting the method across Latin America.

Born from the AI & Equality course and built with CENIA and the University of Chile as the first free course of its kind in the region. (In Spanish.)

The Course

Learning the foundations, free and evergreen.

The original five-module course on a human rights-based approach to AI, with a certificate from the Sorbonne Centre for AI.