A farmer in rural India takes photos of dying plants. She wants to research on the internet, but she doesn’t speak English. She doesn’t have to.
This is the type of problem a nonprofit called Current AI is trying to solve by building an open, public AI infrastructure. We collaborated with Bhashini, the Indian government’s AI language department, at the India AI Summit last February. The result is the Hindi Suno Sutra, a pocket-sized offline device that runs AI in 22 Indian languages without the need for the internet. “There are hundreds of different languages and dialects in India, and AI currently does not represent them,” AI CEO Ayah Bdeir told TechCrunch. The device is open source and can be built by the developer community.
The non-profit organization, founded in February 2025 by Martin Tisne, is moving quickly. Last month, we allocated $3.2 million in grants to projects from four organizations. Most recently (last week) we launched an open source AI chatbot at the AI for Good Summit in Geneva.
Bdeir joined in January after leading Mozilla’s AI strategy. She previously founded littleBits, a STEM education company that reached millions of children, which she sold to Sphero in 2019.
Currently, AI operates as a “public-private partnership,” bringing together governments, businesses and philanthropies to fund technologies for the public good, she told TechCrunch. The French government invested $100 million in Current AI, along with the Ford Foundation, MacArthur Foundation, DeepMind, and Salesforce, bringing total committed funding to $400 million. “They are funders, not investors,” Bdeir said..
The problem I’m trying to solve is simple. From OpenAI to Google to Anthropic, all of today’s major AI systems belong to private companies. “For AI to be a truly transformative technology and transform every aspect of everyone’s life, there must be a public alternative,” Bdeir said. “Like the World Wide Web, it is free for everyone.”
Half of the world’s spoken languages are at risk of extinction. “And with English leading the largest language models and AI systems, most of the world’s languages, and consequently their cultures and communities, have been left behind,” Bdeir said.
When asked about Big Tech’s multilingual push, Bdeir drew a sharp distinction. “Big tech companies build multilingual models to expand their markets, regardless of consent or context,” she said. The results are concrete. “For indigenous languages, missionary Bible translations become training data before the community sets the rules,” she said.
It’s not just about language
An AI’s ability to speak a language is only part of what it needs to learn. “Language is how knowledge, traditions, memories and identity are passed down from one generation to the next. So if technology can’t speak your language, it can’t capture your culture,” she said.
Her vision for Current AI is an open system modeled after the early web, where improvements benefit everyone, no one is blocked, and communities remain in control of their own data.
Current’s first round of cohort grants, announced last month, included distributing $3.2 million to four organizations across the Amazon in Kenya, Lebanon, and Brazil.
Masakane, Kenya’s project involves building AI datasets in more than 50 African languages for health, agriculture and education. Lebanon’s Institute for Worldmaking is digitizing Arab cultural history and contemporary practices into a machine-readable database managed by the community rather than a technology company. In Brazil, Portal sem Porteiras is building offline AI tools with indigenous communities in the Amazon to maintain data within the region. And in Kenya, the African Internet Rights Alliance is developing audit tools to hold AI systems accountable across the continent.
Who owns the data?
On the issue of data ownership, Bdeir did not mince words. “There are many different models and proposals for who owns data in different communities, but one thing is certain: companies in Silicon Valley should not be trying to make certain thousands of people richer,” she told TechCrunch.
The nonprofit’s approach is to store models and data locally, bring in community experts before building anything, or write consent protocols into the pipeline so that the community can stop the process at any time.
None of the beneficiaries of Current AI have fully solved this problem yet. But Bdeir thinks that’s the point. “They all built questions into their work, rather than accepting the common default where complexity is an excuse to let governments or tech companies decide for everyone,” she said.
“Size is not always the metric. This is the Big Tech paradigm,” Bdeir said of how much progress could be made by splitting the $3.2 million budget between four organizations. “It might seem like an indigenous elder in the Brazilian Amazon could use tools made in Kenya to convey ecological knowledge in their own language.”
Building a Stack
Earlier this month in Geneva, Current launched Alpha Chat, an open source chatbot assembled in seven weeks by a coalition of 10 organizations including Hugging Face, Mozilla and MIT Media Lab. Each contributor brought their part of the stack, including language models, safety tools, and computing power.
Current AI also signed a deal with Sakana AI, a Tokyo-based startup known for its work on something called Sovereign AI. The two organizations plan to build a shared open source AI stack designed to support Japanese language and culture, but also to support communities across the Global South that have been largely ignored by dominant AI systems.
If you purchase through links in our articles, we may receive a small commission. This does not affect our editorial independence.
