Key Takeaways

  • A nonprofit backed by $400 million in public and philanthropic funding is building an open, offline AI that runs in 22 Indian languages — no internet required
  • Current AI argues every major AI system today is privately owned; if AI transforms every life, a public alternative must exist, modeled on the World Wide Web
  • Half the world's spoken languages face extinction while English-dominated models extract missionary Bible translations as training data before Indigenous communities set any rules
  • The organization moves like a startup: $3.2 million in grants last month, an open-source chatbot launched in Geneva, and a pocket device co-built with India's government language division

A farmer in rural India photographs a dying crop. She needs answers but doesn't speak English. The internet shrugs. That gap — between the languages AI serves and the languages people actually speak — is where Current AI plants its flag.

The nonprofit, barely six months old, has already secured $400 million in committed funding. The French government led with $100 million. The Ford Foundation, MacArthur Foundation, DeepMind, and Salesforce followed. They are not investors. They are funders. No equity, no board seats, no exit strategy. That distinction matters. It changes what gets built and who it serves.

In February at the India AI Summit, Current AI partnered with Bhashini, the Indian government's language division. They produced Suno Sutra — Hindi for "listening chronicles" — a pocket-sized device that runs AI in 22 Indian languages offline. No connectivity required. The code is open-sourced. Developers can build on it tomorrow. This is not a pilot. It is infrastructure.

Last month the organization deployed $3.2 million in grants across four organizations. Last week it launched an open-source chatbot at the AI for Good Summit in Geneva. Speed is deliberate. Martin Tisne founded Current AI in February 2025. Ayah Bdeir joined as CEO in January after leading Mozilla's AI strategy. She previously founded littleBits, a STEM education company that reached millions of children before selling to Sphero. She knows how to scale mission-driven technology.

The thesis is straightforward: every major AI system — OpenAI, Google, Anthropic — belongs to a private company. If AI truly transforms every aspect of every life, a public alternative must exist. Like the World Wide Web. Available to anyone. Free.

Half the world's spoken languages face extinction. English drives the largest models. The consequence is not merely linguistic. It is cultural erasure. When a technology cannot speak your language, it cannot hold your culture either. Language carries knowledge, tradition, memory, identity from one generation to the next. An AI that ignores a language severs that chain.

Big Tech builds multilingual models to expand markets. Consent and context are afterthoughts. For Indigenous languages, missionary Bible translations become training data before communities have set any rules. Extraction precedes permission. Current AI draws a sharp line there. The communities decide.

The organization operates as a public-private partnership — governments, companies, philanthropies funding public-interest tech together. The model resembles the early web: improvements benefit everyone, no one gets locked out. That architecture produced the internet's most democratic era. Current AI bets it can do the same for intelligence.

Skeptics will ask whether $400 million suffices against the billions Big Tech burns weekly. The question mistakes the game. Current AI does not need to outspend Google. It needs to build the layer Google cannot own — the commons. The web did not beat proprietary networks by outspending them. It beat them by being open.

The farmer in rural India still needs her answer. Today she might wait for a signal that never comes. Tomorrow she could hold a device that speaks her language, understands her context, and respects her community's rules. That is the wager. The web proved open wins. Current AI is betting intelligence will too.