A farmer in rural India photographs a dying plant and wants to look it up online, but she does not speak English. That is the kind of gap Current AI, a nonprofit building open and publicly funded AI infrastructure, says it exists to close — and it now has more than $400 million (around ₱24.6 billion) in committed funding to try.
The organisation's chief executive, Ayah Bdeir, laid out the plan in an interview with TechCrunch published on 19 July. Her framing is deliberately old-fashioned: "If AI is truly a transformative technology, if it's going to change every aspect of everyone's life, there has to be a public alternative," she said. "Like the World Wide Web, available to anyone, for free."
Who funds Current AI, and how much
Current AI was founded in February 2025 by Martin Tisne at the AI Action Summit in Paris. Bdeir joined in January 2026 after leading Mozilla's AI strategy; she earlier founded littleBits, the STEM education company sold to Sphero in 2019.
The money picture, using an exchange rate of roughly ₱61.60 to the US dollar:
| Item | Amount | Approximate peso value |
|---|
| French government seed funding | $100 million | around ₱6.2 billion |
| Total committed to date | $400 million | around ₱24.6 billion |
| Target being mobilised over five years | $2.5 billion | around ₱154 billion |
| First grant round (June 2026) | $3.2 million | around ₱197 million |
Beyond the French government, the core backers include the Ford Foundation, the MacArthur Foundation, DeepMind and Salesforce. Bdeir is careful about the label: "They're not investors; they're funders." On its own site, Current AI describes itself as mobilising $2.5 billion over five years from a coalition of governments, foundations and private-sector partners, anchored by the Paris Charter on Artificial Intelligence in the Public Interest.
Ten countries have signed that charter and committed to contribute: Chile, Finland, France, Germany, India, Kenya, Morocco, Nigeria, Slovenia and Switzerland. The Philippines is not on that list.
What it has actually shipped
The nonprofit has moved quickly for its size:
- Suno Sutra (February 2026) — built with Bhashini, the Indian government's AI language division, and named with the Hindi phrase for "listening chronicles." It is a pocket-sized, offline device that runs AI in 22 Indian languages with no internet connection required, and it is open-sourced so developer communities can build on it.
- First grant cohort (June 2026) — $3.2 million (around ₱197 million) split across four organisations. Masakhane in Kenya is building AI datasets across more than 50 African languages for health, farming and education. Lebanon's Institute for Worldmaking is digitising Arab cultural history into machine-readable databases that communities, not tech companies, control. Brazil's Portal sem Porteiras is building offline AI tools with Indigenous Amazon communities, keeping the data inside the territory. Kenya's African Internet Rights Alliance is developing audit tools to hold AI systems accountable.
- Alpha Chat (July 2026) — an open-source chatbot assembled in seven weeks by a coalition of ten organisations including Hugging Face, Mozilla and the MIT Media Lab, launched at the AI for Good Summit in Geneva. Each contributor supplied part of the stack: a language model, safety tooling, computing power.
- Sakana AI partnership — a deal with the Tokyo startup to build a shared open-source AI stack for the Japanese language and, the two say, for communities across the Global South that dominant AI systems have largely ignored.
Language, and the fight over who owns the data
Bdeir's argument is that language coverage is not a feature but a matter of cultural survival. Half the world's spoken languages face extinction, and, she said, "with English driving the largest language models and AI systems, a bulk of the world's languages and, consequently, cultures and communities are left behind."
She draws a sharp line between that work and Big Tech's multilingual efforts. "Big tech builds multilingual models to expand their market," she said, "regardless of consent or context." Her example is concrete: "For Indigenous languages, missionary Bible translations become training data before communities have set any rules."
On ownership, she was direct: "There are different models and proposals for who owns data in various communities, but one thing is sure: it shouldn't be a company in Silicon Valley trying to make a select few thousand people wealthier." Current AI's method is to store models and data locally, bring community experts in before anything is built, and write consent protocols into the pipeline so a community can stop the process at any point. None of its grantees has fully solved the ownership question, Bdeir conceded — but "every one of them has built the question into their work."
Asked whether $3.2 million across four organisations can achieve much, she pushed back on the premise: "Scale is not always the measure. That is the Big Tech paradigm."
Where this leaves Filipino languages
The Philippines has roughly 180 living languages, and only a handful of them are well represented in the datasets that train large AI models. Tagalog gets reasonable coverage; Cebuano, Ilocano, Hiligaynon, Waray and dozens of smaller languages get much less. Current AI's model — offline devices, locally held data, community consent — maps neatly onto that situation, and its Suno Sutra device for 22 Indian languages is a working example of the approach.
But there is no Philippine project, grant or partner country announcement from Current AI as of publication. Filipino researchers and community groups would have to enter through its open grant and coalition process, and the tools it has released so far, including Alpha Chat and Suno Sutra, are open-source and free to build on. It is a similar pattern to the one we saw when China launched its WAICO AI alliance without the Philippines among its founding members: the international AI-governance tables are filling up, and Philippine institutions are not yet sitting at them.
The honest limits
Current AI's committed $400 million is real money, but it is small next to the tens of billions that single private labs spend on training runs and data centres each year. Its bet is that coordination, open licensing and community consent can go further than raw compute. Whether that holds up is not something a single funding announcement can settle — the test will be whether Alpha Chat, Suno Sutra and the grantee projects are still being used and improved a year from now.
