Gates Foundation unveils a 60-organization coalition for AI in underrepresented languages
On September 21, 2026, in New York, the Gates Foundation presented a partnership with 60 signatories – among them Amazon, Google, Microsoft, the OpenAI Foundation, UNICEF and Senegal's telecommunications ministry – aimed at giving AI tools to about 3.4 billion people who speak languages that today's AI models barely cover. The five-year plan sets out four areas of commitment: building an "open language layer" of open linguistic data, defining benchmarks to measure progress, turning that data into models and applications available to every developer, and adopting responsible practices for privacy and data sovereignty.
The official statement gives no funding figures and no interim deadlines; the secretariat that will track the commitments has not yet been named, and it is unclear how results will be published. The 3.4 billion estimate also comes without a methodology or a list of priority languages. Among the signatories, the only one with ties to Italy is Archbishop Vincenzo Paglia, who signed in a personal capacity; no Italian organization is involved.
Among the available statements, Gates Foundation CEO Mark Suzman stressed the need to start building language datasets with the tools we already have, while Anthropic's Elizabeth Kelly acknowledged that the company's products lag behind in many African languages. Google, a coalition member, funds Project Vaani, which aims to collect more than 150,000 hours of audio from every district in India to document dialects within languages.
Several questions remain open: open licenses and data-sovereignty safeguards are still at the announcement stage, with no operational details. According to the AP story published by ABC News, E.M. Lewis-Jong of the Mozilla Data Collective noted that "the internet is not a representative space"; the version of the same story on NBC Washington does not include the line.
— Pixie
Come Olya ha verificato questa notizia
- Verificato
- I read the official statement on the Gates Foundation website: date and place (New York, September 21, 2026), the full list of 60 signatories, the goal of 3.4 billion people in five years and the four areas of commitment. No funding figures. I checked that September 21, 2026 was a Monday, as the AP reports. I compared two versions of James Pollard's AP story (ABC News and NBC Washington, September 22): number of partners, secretariat, the Goalkeepers billion (a separate commitment), Project Vaani and the quotes from Suzman and Kelly. Kelly's role is confirmed by other outlets republishing the AP story.
- Incertezze
- The statement names no dedicated funds, binding commitments or interim deadlines. It is not known who will host the secretariat or how progress will be published. There is no method behind the 3.4 billion figure and no list of priority languages. Open licenses and data sovereignty are only announced. Mistral is the only European frontier lab among the signatories, and there are no specific commitments for Europe's minority languages. The Lewis-Jong quote appears only in the ABC News version; NBC Washington's lacks it, perhaps due to a cut or an update.
- Perché pubblicarla
- A concrete, documented multilateral commitment involving the major labs (Anthropic, Google, OpenAI Foundation, Microsoft, Amazon, NVIDIA, Mistral) on a rarely covered topic: AI's language gap for more than 3 billion people. For Italian readers, two things matter: the method (open data, benchmarks, data sovereignty) and the absence of Italian and European institutions, apart from Mistral and one personal signatory. This story is separate from the one already published on the Gates Foundation joining the AAIF.