Whose Knowledge? and the Distributed AI Research Institute (DAIR) are hosting a virtual launch of their joint research on Community-Centered Sovereign Language Technologies on Thursday, November 12, 2026, 15:30 UTC. The event is free to attend online, and registration is open on Zoom. The organisers say mainstream AI development is still dominated by extractive models, and their research-in-action initiative asks a different question: what happens when communities build language technology themselves, guided by collective stewardship, local values, and lived experience? The work focuses on East and West African communities, with a particular emphasis on Kiswahili and Tigrinya.

At the launch, the team will share their findings through an in-depth report, a new storytelling platform, and best-practice guidelines for building community-centered language technologies, among other resources. DAIR describes the underlying project as an effort to develop “sovereign” language technologies for East African languages, using community-centered design and governance practices that help communities create and adapt content for online knowledge repositories. In practice, that means communities have a say over how their languages are collected, used, and governed in AI systems, rather than having their data scraped by someone else.

The stakes are especially high for languages like Tigrinya, spoken by millions across Eritrea and Ethiopia’s Tigray region but still badly under-served by language technology, as researchers have noted. DAIR was founded by computer scientist Timnit Gebru, and the organisation has worked with partners on the continent, including the Huniki Federation, to build tools for languages that big tech tends to ignore. Whose Knowledge?, a global campaign to centre the knowledge of communities that have been marginalised online, runs a language justice programme and has described its work with DAIR and Tigrinya speakers on its podcast. Asmelash Teka Hadgu, a DAIR fellow and the founder of Lesan AI, has spoken about the difference between using the internet in English and in his native Tigrinya.

For Brittle Paper readers, the questions the event raises go beyond technology. Who gets to write, publish, and be read online in an African language? Who owns the data that is used to train a model on Kiswahili or Tigrinya poetry? As debates about AI and creative work continue to grow, with writers and authors’ organisations working out where they stand, this conversation about language, ownership, and community control is an important one. If you work with African languages as a writer, translator, publisher, or researcher, it’s worth attending. Register for the launch here.