Students learning together in a classroom in a low income country
Education

GAILA Unites UNICEF, Google and Anthropic to Bring AI Learning to the World's Poorest Classrooms

7 min read|Updated November 2026
Share

At a glance

GAILA is a global coalition launched in February 2026 led jointly by UNICEF, with its secretariat run by Dalberg and initial support from the Gates Foundation. It brings frontier AI companies including Google and Anthropic together with governments and education organizations to make AI learning tools genuinely usable in low and middle income countries, where most of the world's schoolchildren live but almost none of the AI investment goes. More than 40 organizations have endorsed its Call to Action, which commits them to shared public goods, better model performance for these countries, stronger safety standards, and lower costs.

  • Launched on February 17, 2026 on the sidelines of the India AI Impact Summit in New Delhi
  • Co-led by UNICEF with the secretariat operated by Dalberg and initial support from the Gates Foundation
  • More than 40 organizations have endorsed the AI for Learning Call to Action, including Anthropic, BRAC, Pratham and EdTech Hub
  • An Industry Compact with measurable commitments on performance, safety, accessibility, affordability and relevance is being developed jointly during 2026
  • Participants include the Governments of India, Kenya and Rwanda, plus Google, Anthropic and Karya AI

The Global AI for Learning Alliance, known as GAILA, is a new international coalition that wants to make sure artificial intelligence improves learning for children in low and middle income countries, not just wealthy ones. Launched on February 17, 2026 on the sidelines of the India AI Impact Summit in New Delhi, the alliance coordinates AI firms, governments and education organizations around a shared problem: frontier AI investment is concentrated in higher income countries while the countries with the largest school age populations face limited infrastructure, affordability constraints and tools that were never designed for their classrooms.

The initiative matters because it represents the first serious attempt to fix a structural imbalance in AI and education. Instead of every company and nonprofit acting alone, GAILA aligns incentives, pools investment into shared public goods, and holds participants to public commitments on safety, affordability and relevance. The alliance is led jointly by UNICEF and partners, its secretariat is operated by the global advisory firm Dalberg, and it has initial support from the Gates Foundation.

Why AI Education Tools Fail in the Countries That Need Them Most

Most large language models are trained overwhelmingly on English text written in wealthy countries. When a student in rural Kenya or a teacher in Bihar asks a question in Kiswahili, Hindi or Sheng, the models frequently misunderstand, answer incorrectly or fall back to generic responses. On top of language gaps, schools in low and middle income countries face patchy internet access, shared devices, large class sizes and teachers who are themselves under supported.

The result is a growing risk that AI accelerates educational inequality instead of reducing it. Pia Britto, Global Director of Education and Adolescent Development at UNICEF, framed the mission at the launch event. GAILA, she said, is designed to go beyond dialogue and serve as a bridge from commitment to implementation, translating innovation into measurable improvements in learning outcomes.

Who Has Joined the Alliance

To date, more than 40 organizations have endorsed the AI for Learning Call to Action. The list includes Anthropic and Google alongside major education and development actors such as BRAC, Pratham, the Central Square Foundation, EkStep Foundation, EdTech Hub, Education Above All, and the United Kingdom's FCDO. Governments are at the table too. India's Ministry of Education, the Government of Kenya and Rwanda all took part in the launch, which was hosted by the Observer Research Foundation together with Carnegie India (gaila) and Carnegie India.

The private sector participation is what makes GAILA unusual. Elizabeth Kelly, head of beneficial deployments at Anthropic, said at the launch that her company is already seeing what happens when AI is built with communities rather than deployed to them, citing partnerships with governments, educators and nonprofits in Rwanda and India. A joint statement from Google and Google DeepMind stressed that getting AI in education right requires cross sector collaboration, with tools grounded in learning science and distributed equitably. Karya AI, the Indian nonprofit that pays rural communities for data work, also participated.

What Participants Are Actually Committing To

GAILA's starting point is the AI for Learning Industry Call to Action, which outlines joint commitments across four areas. First, coordinated investment in shared public goods, so that tools, datasets and evaluations are built once and reused rather than duplicated. Second, improved model performance for education in low and middle income countries, meaning models that handle local languages, curricula and real classroom conditions. Third, strengthened standards for safety, ethics and contextual relevance. Fourth, expanded affordability and uptake of AI models, so cost stops being the barrier it is today.

During 2026, the alliance is developing a fuller Industry Compact with specific, measurable and time bound commitments, including benchmarks for learning results and privacy rules that work across different countries. Benjamin Piper, director of global education at the Gates Foundation, said at the launch that the ecosystem needs to work better and expressed excitement about the progress made in the room.

The Global South Setting Its Own Rules

A defining feature of GAILA is that low and middle income countries are joining as rule makers, not just customers. Paula Ingabire, Rwanda's Minister for Communication Technology and Innovation, said that by joining GAILA Rwanda is joining both a vision and a community, because the Global South must write the rules of how AI serves learning rather than simply adopting tools developed elsewhere. Rohit Tripathi, deputy secretary at India's Ministry of Education, described two national mandates: building the talent pipeline and using AI to deliver better educational services such as personalized learning.

The momentum has continued beyond the launch. In July 2026, Google announced new initiatives supporting India and the African Union Commission to help millions of teachers and students build AI literacy, the kind of national scale effort GAILA exists to coordinate. Pilot programs are already running in member countries, including AI teaching assistants deployed in Rwandan classrooms through the EdTech Hub sandbox program.

Why Coordination Beats Fragmentation

Education technology has a long history of failed pilots: well funded tools that work in one district and never scale, duplicated efforts that drain the same donor pool, and products that ignore what teachers actually need. GAILA's theory of change is that these failures are collective action problems, not technology problems. By convening AI firms, governments, researchers and frontline educators under shared commitments, the alliance aims to reduce fragmentation and direct attention to national priorities.

If the 2026 Industry Compact delivers on its promise of measurable commitments, the world's largest student populations could get AI tutors, teacher assistants and adaptive learning tools designed for their languages, their curricula and their budgets. That would mark a genuine shift: from AI education tools built for the wealthy world and adapted downward, to tools built with the Global South from the start.

Common Questions

What is GAILA?

GAILA, the Global AI for Learning Alliance, is a coalition launched in February 2026 that brings together AI firms, governments and education organizations to scale effective and responsible AI enabled learning in low and middle income countries. It is led jointly by UNICEF and partners, with the secretariat operated by Dalberg and initial support from the Gates Foundation.

When and where was the Global AI for Learning Alliance launched?

GAILA was formally launched on February 17, 2026 on the sidelines of the India AI Impact Summit in New Delhi. The launch event was hosted by the Observer Research Foundation together with Carnegie India (gaila) and Carnegie India.

Which companies and organizations have joined GAILA?

More than 40 organizations have endorsed the AI for Learning Call to Action, including Anthropic, Google, BRAC, Pratham, the Central Square Foundation, EkStep Foundation, EdTech Hub, Education Above All, Dalberg, FCDO and the Gates Foundation. The Governments of India, Kenya and Rwanda also participated in the launch.

What commitments do GAILA members make?

The Call to Action outlines four commitment areas: coordinated investment in shared public goods, improved model performance for education in low and middle income countries, strengthened standards for safety, ethics and contextual relevance, and expanded affordability and uptake of AI models. A fuller Industry Compact with measurable, time bound commitments is being developed jointly during 2026.

Why does GAILA focus on low and middle income countries?

Frontier AI investment remains concentrated in higher income countries, while education systems in low and middle income countries face limited infrastructure, affordability constraints and weak alignment between technology and system needs. Most of the world's school age children live in these countries, so GAILA aims to ensure AI narrows the learning gap rather than widening it.