Current AI: Democratizing AI Without Language Barriers

An Initiative to Overcome Language Barriers

In a remote region of India, a farmer faces a common problem: she wants to identify a disease affecting her plants, but the language barrier prevents her from accessing online resources. This situation illustrates the challenge that Current AI, a nonprofit organization, seeks to address by developing an artificial intelligence (AI) infrastructure that is accessible to all, regardless of spoken languages.

At the India AI Summit in February, Current AI announced a partnership with Bhashini, a division of the Indian government dedicated to languages. Together, they created Suno Sutra, a portable device that works offline and uses AI to process 22 Indian languages. This fully open-source device is designed for use by community developers, bridging the language gap that prevents many people from accessing modern technologies.

Ambitious Funding for Global Projects

Founded in February 2025 by Martin Tisne, Current AI has quickly gained traction. Last month, the organization awarded $3.2 million in grants to four projects across different regions of the world. More recently, it unveiled an open-source AI chatbot at the AI for Good Summit in Geneva.

Ayah Bdeir, who joined Current AI in January after leading Mozilla's AI strategy, is behind this momentum. She is also known for founding littleBits, an educational STEM company that reached millions of children before being sold to Sphero in 2019.

Current AI operates as a public-private partnership, bringing together governments, businesses, and foundations to fund public interest technologies. The French government initially contributed $100 million, followed by the Ford Foundation, the MacArthur Foundation, DeepMind, and Salesforce, bringing the total funding to $400 million. Bdeir clarifies that these funds are not investments but donations.

A Public Alternative to AI Giants

Bdeir highlights a fundamental issue: current AI systems, whether developed by OpenAI, Google, or Anthropic, all belong to private companies. She asserts that if AI is a truly transformative technology, it is crucial to provide a public alternative, akin to the World Wide Web, accessible to all at no cost.

Linguistic diversity is also at risk, with half of the world's languages threatened with extinction. English dominates language models and AI systems, sidelining many cultures and communities. Bdeir criticizes large tech companies that develop multilingual models primarily to expand their market, often without considering consent or cultural context.

Preserving Cultures Through Technology

The ability of an AI to understand a language is just one aspect of its potential. Language is a vehicle for transmitting knowledge, tradition, and cultural identity. If a technology cannot communicate in your language, it cannot preserve your culture, Bdeir explains.

She envisions an open system for Current AI, inspired by the early days of the web, where improvements benefit everyone and communities retain control over their data. The first grant cycle of Current AI has funded projects in Kenya, Lebanon, and the Brazilian Amazon.

Concrete Projects for Global Impact

In Kenya, the Masakhane project aims to create AI datasets in over 50 African languages, covering areas such as health, agriculture, and education. In Lebanon, the Worldmaking Institute is digitizing Arab cultural history in databases controlled by local communities. In Brazil, Portal sem Porteiras is developing offline AI tools with Indigenous Amazonian communities, ensuring that data remains on their territory. Finally, the African Internet Rights Alliance in Kenya is developing audit tools to hold AI systems accountable on the continent.

The Crucial Question of Data Ownership

Bdeir is clear on the issue of data ownership: it should not be in the hands of a Silicon Valley company seeking to enrich a minority. Current AI's approach is to store models and data locally, consulting community experts before any construction, and integrating consent protocols so that communities can halt the process at any time.

Although Current AI's beneficiaries have not yet resolved this issue, Bdeir sees it as an advantage. Each project incorporates this question into its work, refusing to conform to the usual model where complexity justifies decisions being made by governments or tech companies.

A Model of Collaborative Innovation

In Geneva, Current AI recently launched Alpha Chat, an open-source chatbot developed in seven weeks through a coalition of ten organizations, including Hugging Face, Mozilla, and the MIT Media Lab. Each partner contributed to the infrastructure, providing language models, security tools, and computing power.

Finally, Current AI has signed an agreement with Sakana AI, a Tokyo-based startup specializing in sovereign AI. Together, they plan to build a shared open-source AI infrastructure aimed at supporting the Japanese language and culture, as well as communities in the Global South often overlooked by dominant AI systems.