In an era increasingly shaped by the powerful algorithms of artificial intelligence, a fundamental question emerges: who controls this transformative technology, and for whose benefit? This query lies at the heart of Current AI, a pioneering non-profit organization dedicated to constructing open, public AI infrastructure, envisioned as a "World Wide Web of AI" accessible and beneficial to all. Its mission directly confronts the prevailing paradigm where leading AI systems are largely proprietary, developed by private corporations, and predominantly trained on data reflecting a narrow linguistic and cultural spectrum.

The Digital Language Divide and the Urgency of Inclusion

The genesis of Current AI’s ambitious endeavor stems from a profound observation: the current trajectory of AI development risks exacerbating global inequalities, particularly for the billions who do not speak dominant languages like English. Imagine a farmer in a remote Indian village, her livelihood threatened by a dying crop. She seeks information, perhaps by photographing the ailing plant, but her native language isn’t English, nor is reliable internet access a given. In the present AI landscape, her query might go unanswered, or worse, be poorly interpreted through culturally insensitive translations. This scenario epitomizes the digital language divide—a chasm that Current AI is determined to bridge.

The challenge is stark. According to the United Nations, approximately half of the world’s spoken languages face extinction by the end of this century. While the UN has declared 2022-2032 as the International Decade of Indigenous Languages to highlight this crisis, the rapid advancement of AI, heavily reliant on vast datasets, threatens to accelerate this loss if left unchecked. Large language models (LLMs) from tech giants, driven by commercial imperatives, prioritize market expansion. This often means developing multilingual capabilities by scraping data indiscriminately, frequently without the consent or contextual understanding of the communities whose linguistic heritage is being digitized. For vulnerable Indigenous languages, this can mean missionary Bible translations becoming foundational training data before communities can even establish their own digital sovereignty protocols, as Ayah Bdeir, CEO of Current AI, has pointed out. This not only risks misrepresentation but also the loss of unique cultural nuances embedded within language itself.

A Vision for Public AI: Learning from the Web’s Open Spirit

Current AI, founded in February 2025 by Martin Tisne, explicitly draws inspiration from the early internet. The World Wide Web, initially conceived as an open, decentralized network for information sharing, became a public utility, fostering unprecedented global connectivity and innovation precisely because its foundational protocols were open and accessible. Current AI aims to replicate this spirit for artificial intelligence, establishing a public alternative to the closed, proprietary systems that currently dominate the AI landscape.

Ayah Bdeir, who joined Current AI in January after spearheading Mozilla’s AI strategy and previously founding littleBits, an educational technology company, articulates this vision clearly. If AI is poised to fundamentally reshape every aspect of human life, she argues, then a public, open-source alternative is not merely desirable but essential. This commitment to openness extends beyond technical specifications to encompass ethical principles, data governance, and community empowerment. The goal is to ensure that the benefits of AI are distributed equitably, fostering an environment where innovation can flourish without imposing a singular cultural or linguistic worldview.

The Operational Model: Public-Private Partnerships for Public Good

Current AI operates on a unique "public-private partnership" model, strategically pooling resources from governments, corporations, and philanthropic organizations to fund public interest technology initiatives. This collaborative approach recognizes the immense capital and expertise required to build robust AI infrastructure while safeguarding it from purely commercial pressures. The French government provided initial seed funding of $100 million, a significant commitment that quickly attracted further support. Prestigious entities like the Ford Foundation, the MacArthur Foundation, DeepMind, and Salesforce have since joined, bringing the total committed funding to $400 million. Bdeir emphasizes that these contributors are "funders," not investors, signifying a shared commitment to the public good rather than a pursuit of financial returns. This distinction is crucial, allowing Current AI to prioritize long-term societal impact over short-term commercial viability.

Pioneering Initiatives: From Offline Devices to Open-Source Chatbots

Current AI’s strategy involves a multi-pronged approach, encompassing both foundational infrastructure development and targeted projects that address immediate needs. One of its earliest and most tangible successes is Suno Sutra, which translates to "listening chronicles" in Hindi. Developed in collaboration with Bhashini, the Indian government’s AI language division, and unveiled at the India AI Summit in February, Suno Sutra is a groundbreaking pocket-sized, offline device. It runs AI in 22 Indian languages, entirely independent of internet connectivity. This innovation is particularly vital in a country like India, with its hundreds of diverse languages and dialects, many of which are underserved by mainstream AI. By being open-sourced, Suno Sutra empowers developer communities to build upon its foundation, tailoring it to even more specific linguistic and cultural contexts.

Beyond localized hardware, Current AI is also building an open software stack. Last week, at the AI for Good Summit in Geneva, the organization launched Alpha Chat, an open-source chatbot assembled in just seven weeks by a coalition of ten diverse organizations, including Hugging Face, Mozilla, and MIT Media Lab. This rapid development demonstrates the power of collaborative, open-source models, where each contributor brings a specialized piece of the puzzle—be it a language model, safety tooling, or computing power—to create a functional and accessible tool. This initiative directly counters the proprietary nature of dominant chatbots, offering a transparent and customizable alternative.

Further solidifying its technical backbone, Current AI recently forged a strategic alliance with Sakana AI, a Tokyo-based startup renowned for its "Sovereign AI" research. This partnership aims to jointly construct a shared open-source AI stack, initially designed to support the Japanese language and culture. Crucially, this stack will also be engineered to serve communities across the Global South that have historically been overlooked by prevailing AI systems. The concept of "Sovereign AI" aligns perfectly with Current AI’s ethos, emphasizing the development of AI that is culturally specific, locally controlled, and respectful of national and community data governance.

Empowering Local Communities: A Grant-Making Paradigm

Current AI’s commitment to localized impact is perhaps best exemplified by its grant program. Last month, it allocated $3.2 million to a diverse cohort of four organizations across Kenya, Lebanon, and the Brazilian Amazon, demonstrating a clear focus on regions often marginalized in global tech development. These grants are not merely financial contributions but strategic investments in building capacity and fostering self-determination within these communities.

In Kenya, the grant supports Masakhane, an initiative focused on building extensive AI datasets across more than 50 African languages, specifically tailored for applications in health, farming, and education. This directly addresses the critical need for locally relevant data, without which AI models cannot effectively serve these communities. Simultaneously, the African Internet Rights Alliance, also in Kenya, is developing crucial audit tools to hold AI systems accountable across the continent, advocating for ethical AI deployment and data protection.

In Lebanon, the Institute for Worldmaking is utilizing its grant to digitize Arab cultural history and contemporary practices into machine-readable databases. Critically, these databases are designed to be controlled by the communities themselves, not by external tech corporations, ensuring cultural ownership and preservation. Similarly, in the Brazilian Amazon, Portal sem Porteiras is working directly with Indigenous communities to develop offline AI tools, maintaining data within territorial boundaries and respecting traditional knowledge systems. This approach directly challenges the extractive models often seen in technology, where valuable data is siphoned off without adequate consent or benefit to its originators.

The Crux of Data Ownership: Community Control and Consent

The question of data ownership is central to Current AI’s philosophy and operations. Ayah Bdeir minces no words on the subject: "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." This firm stance underpins Current AI’s approach, which prioritizes local data storage, community expert involvement from the earliest stages of development, and the integration of robust consent protocols into every pipeline. These protocols empower communities to halt or modify processes at any point, ensuring genuine agency over their digital assets.

While acknowledging that none of its grantees have fully "solved" the complex issue of data ownership, Bdeir views this ongoing exploration as a fundamental strength. By embedding this question into the very fabric of their work, Current AI’s partners are actively resisting the default assumption that technological complexity justifies external control. Instead, they are pioneering new models where communities, rather than governments or corporations, are the ultimate arbiters of their data and digital future.

Challenges and the Path Forward: A Different Measure of Success

Building a truly open, public, and equitable AI infrastructure is an undertaking fraught with challenges. Technical hurdles, the immense computational resources required, and the complexities of governance across diverse linguistic and cultural contexts are significant. Furthermore, securing sustained funding for public interest technology often competes with the lucrative, venture-backed world of proprietary AI. However, Current AI consciously rejects the "Big Tech paradigm" that equates success solely with scale and market dominance.

"Scale is not always the measure," Bdeir asserts. Instead, Current AI envisions a different kind of impact, one measured by empowerment and cultural resilience. This could manifest as an Indigenous elder in the Brazilian Amazon utilizing a tool, perhaps even one built in Kenya, to transmit vital ecological knowledge in their ancestral language to future generations. Such an outcome, though seemingly small in a global context, represents a profound victory for cultural preservation and equitable access to technology.

Current AI’s journey is just beginning, but its rapid progress and strategic partnerships signal a significant shift in the global AI landscape. By championing openness, prioritizing cultural context, and empowering local communities, Current AI is not merely building new technologies; it is forging an inclusive digital frontier, laying the groundwork for an artificial intelligence that truly serves humanity in all its rich diversity.