AI language access is the focus of a new five-year commitment by 60 organisations seeking to expand access to AI tools for an estimated 3.4 billion people who speak languages currently underrepresented in AI models.

The organisations, drawn from across the global AI ecosystem, announced the joint commitment on September 21, 2026. The initiative brings together AI developers, researchers, governments, philanthropies, community organisations and other groups working across technology, funding and public policy.

The announcement was published by the Gates Foundation, which is coordinating the joint commitment and has invited additional organisations across the ecosystem to contribute resources and expertise.

Today, only a small percentage of the world’s roughly 7,000 languages are considered sufficiently resourced to support strong AI capabilities. Many languages remain underrepresented in the datasets, tools and benchmarks used to develop and evaluate AI systems.

As a result, speakers of these languages can face systems that are less accurate, less useful or less capable of understanding how they communicate.

The challenge also extends beyond text translation. Voice can be particularly important in communities where typing or text-based interfaces are less practical or accessible. Dialects, slang, idioms and cultural context can also affect how AI systems interpret and respond to users.

The participating organisations aim to address these gaps across four areas.

First, they plan to build an open language layer consisting of shared and safe data infrastructure that AI developers can access through open licences.

Second, they will develop assessments and benchmarks to track progress against the shared goal and measure improvements across underrepresented languages.

Third, the initiative will focus on turning language data into models and applications that can be used by AI developers, including those without access to the resources of larger organisations.

Finally, the organisations will focus on responsible implementation, with privacy, consent and data sovereignty incorporated into the work.

The initiative could support AI-enabled services across sectors including healthcare, education, agriculture, financial services and public services.

The coalition’s initial signatories include organisations such as the Gates Foundation, Amazon, Anthropic, Google, Microsoft, NVIDIA, the OpenAI Foundation, UNICEF and the World Bank Group, alongside research institutions, governments, community organisations and other participants.

However, the commitment is focused on a five-year goal rather than an immediate expansion of AI services across all underrepresented languages. The participating organisations will contribute according to their respective expertise and capabilities, while the coalition’s detailed structure, governance and workstreams are expected to be developed collaboratively over the coming year.

The initiative also builds on existing work across datasets, language models, benchmarks, applications and local implementation. According to the Gates Foundation, progress will continue to be reported as milestones are reached.

The five-year goal is ultimately centred on AI language access, with the participating organisations seeking to enable an estimated 3.4 billion people who speak currently underrepresented languages to use AI tools in their own language and voice.

Related post

View all