Most AI systems are trained on the languages, data, and assumptions of a handful of countries. This summit asks a plainer question: who gets left out, and what would it take to include them properly.
A model available everywhere isn't equitable if it only works well in a handful of languages, or reflects the assumptions of a handful of cultures.
Thousands of languages have little to no representation in today's foundation models — a gap the Compact treats as a governance failure, not a footnote.
Whose data trains the model, and who consents to it, is treated as a first-order question — not a technical afterthought.
Delegates track deployment impact by region and community, not just adoption numbers.
This page is an information page about the AIFOD Geneva Summit 2026. All registration is handled on the official UN Indico platform.
Continue to UN Indico NowRedirecting automatically in 3 seconds…