A farmer in rural India photographing a dying plant but unable to research it online because she does not speak English illustrates the problem Current AI aims to address: building open, public AI infrastructure that works for many languages and can run offline.
What they’ve built so far
At the India AI Summit in February, Current AI partnered with Bhashini, the Indian government’s language unit, to produce Suno Sutra — Hindi for “listening chronicles” — a pocket-sized offline device that runs AI in 22 Indian languages without requiring internet access. The device is open source and available for developer communities to extend.
Recently at the AI for Good Summit in Geneva, Current AI also unveiled an open-source chatbot. Earlier in the month they launched Alpha Chat, an open-source chatbot assembled in seven weeks by a coalition of ten organisations.
Rapid start and funding
Current AI was founded in February 2025 by Martin Tisne. Ayah Bdeir joined as CEO in January; she previously led Mozilla’s AI strategy and founded the STEM education company littleBits, which was acquired by Sphero in 2019.
The organisation describes itself as a public-private partnership: the French government seeded Current AI with $100 million, and donors including the Ford Foundation, the MacArthur Foundation, DeepMind and Salesforce have brought total committed funding to $400 million. Bdeir emphasises these partners are funders rather than investors.
First grant round: $3.2 million
Last month Current AI allocated $3.2 million in grants across four organisations. The funded projects span regions and use cases:
- Masakhane (Kenya): building AI datasets across more than 50 African languages for health, farming and education.
- Institute for Worldmaking (Lebanon): digitising Arab cultural history and contemporary practice into machine-readable databases that communities, not tech companies, control.
- Portal sem Porteiras (Brazilian Amazon): developing offline AI tools with Indigenous Amazonian communities, keeping data within their territory.
- African Internet Rights Alliance (Kenya): creating audit tools to hold AI systems accountable across the continent.
Bdeir notes none of the grantees have fully resolved data ownership issues yet, but each project incorporates the question into its work rather than accepting default solutions.
Why language and community control matter
Current AI’s stated goal is to ensure that AI does not remain concentrated within private, typically English-centric systems. Bdeir argues that if AI is truly transformative and will affect every aspect of life, there must be a public alternative, similar to the early World Wide Web that was freely accessible to anyone.
She stresses that language carries knowledge, tradition, memory and identity between generations — if technology cannot speak a community’s language, it cannot steward its culture. Bdeir also distinguished Current AI’s approach from Big Tech’s multilingual efforts, saying large companies often build multilingual models to expand markets without adequate consent or contextual safeguards. In practice, she warned, some Indigenous languages can end up represented in training data — for example through missionary Bible translations — before communities have set rules about use.
Data ownership and local storage
Current AI’s approach favours storing models and data locally, involving community experts before building anything, and embedding consent protocols into development pipelines so communities can stop the process at any point. Bdeir was explicit that decisions about data and benefit should not be left to a Silicon Valley company seeking to enrich a select few.
International partnerships and a shared stack
Alpha Chat’s assembly involved contributions from organisations such as Hugging Face, Mozilla and MIT Media Lab; each partner provided parts of the stack, including a language model, safety tooling, and compute.
Current AI has also signed a deal with Sakana AI, a Tokyo-based startup known for its work on so-called Sovereign AI. The two organisations plan to build a shared open-source AI stack intended to support Japanese language and culture while also serving communities in the Global South that dominant AI systems have largely ignored.
Practical implications
Current AI envisions an open system similar to the early web: improvements benefit everyone, people are not locked out, and communities retain control over their data and cultural heritage. The organisation argues that progress should not be measured solely by scale as defined by Big Tech; a $3.2 million grant split across four projects can produce meaningful local impact — for example, enabling an Indigenous elder in the Brazilian Amazon to pass ecological knowledge in their own language using a tool developed in Kenya.
Current AI’s activities point to an alternative path for AI development: open, community-driven and locally governed systems that prioritise the preservation of languages and cultures.
