African-built Swahili AI unveiled at Mobile World Congress

An artificial intelligence (AI) model developed specifically for Swahili speakers was presented at the Mobile World Congress (MWC) in Barcelona, marking a significant step towards greater linguistic inclusion in global technology, according to RFI and agencies on March 4th. 

The announcement took place during the 20th edition of the world’s largest telecommunications trade fair, where thousands of industry leaders gather each year to present their innovations.

The model’s development reflects growing efforts to ensure that AI systems serve communities beyond the predominantly English-speaking markets that have traditionally shaped the industry.

Swahili is spoken by millions across East and Central Africa, in countries such as Somalia, Kenya, Tanzania and the Democratic Republic of the Congo.

Despite its widespread use, the Swahili language is underrepresented in major AI systems, mainly due to limited data and resources.

Most widely available language models have been trained primarily on English and other globally dominant languages, limiting accessibility for many African users.

This is despite Africa being home to over 2,000 languages, of which fewer than 3% are represented in AI models.

The project is led by GSMA Africa, a global trade association for mobile network operators that organised the MWC; GSMA estimates that sub-Saharan Africa has over 1.1 billion mobile money accounts. 

Developed through collaboration between African telecom actors and international partners, the model is designed to understand and generate text directly in Swahili.

Its creators argue that offering AI tools in local languages can significantly expand access to digital services and encourage broader adoption of emerging technologies.

Additionally, the initiative seeks to ensure that Swahili speakers can interact with AI in culturally and linguistically appropriate ways.

Project supporters emphasise that language is more than a communication tool, as it also carries social context, nuance and identity, which are essential for building more inclusive digital ecosystems.

Despite Africa often being perceived as a consumer of digital products developed elsewhere, initiatives like this demonstrate growing local participation in the design and production of advanced technologies.

As AI continues to transform industries worldwide, the question of linguistic representation has become more urgent; experts argue that failing to integrate diverse languages risks a deepening of digital inequalities.

By placing Swahili at the centre of a major global technology forum, the project sends a broader message about the importance of diversity in the development of AI systems.

It suggests a shift toward a more multilingual and globally representative vision of artificial intelligence: one in which innovation is shaped by, and accessible to, a wider range of communities.

RFI and agencies, Maghrebi.org