Current AI Seeks to Decouple Global Infrastructure From Silicon Valley Giants

A farmer in rural India photographs a struggling crop. She needs agricultural advice but the internet resources available do not support her native tongue. This gap between the reach of global technology and the reality of local language usage represents the core friction point for Current AI. The nonprofit emerged in February 2025 under the guidance of founder Martin Tisné, with a mandate to construct infrastructure that keeps the power of intelligence systems within the public domain. As TechCrunch reported, the organization operates on the premise that if artificial intelligence is to define the future of human progress, the underlying tools must belong to everyone.

Challenging the Private Corporate Paradigm

CEO Ayah Bdeir joined the organization in January, bringing experience from her previous role leading AI strategy at Mozilla and her tenure as the founder of the education startup littleBits. Bdeir characterizes the current state of technology as a monopoly of convenience, where giants like OpenAI, Google, and Anthropic dictate the terms of access and usage. For the nonprofit, the goal is to shift that power dynamic.

Current AI functions as a public-private partnership. The initiative secured $400 million in funding, with the French government providing a $100 million seed injection. Additional backing came from a coalition of philanthropies and corporations, including the Ford Foundation, the MacArthur Foundation, DeepMind, and Salesforce. Bdeir maintains that these entities act as funders rather than investors, preventing the profit-driven motives that typically constrain open development.

Infrastructure for Endangered Languages

Language acts as the primary vessel for human culture and memory. Current AI estimates that half of the world’s spoken languages face extinction. When dominant systems rely exclusively on English, they inadvertently erase the knowledge and identity of those outside the Anglosphere. The nonprofit seeks to correct this imbalance through localized innovation.

In India, the group teamed up with Bhashini, the government’s dedicated AI language division. Their partnership produced Suno Sutra, a portable, offline device that processes AI tasks across 22 different Indian languages. Because the device functions without an internet connection, it circumvents the infrastructural barriers that often plague rural communities. This open-source hardware allows regional developer communities to customize the stack to suit local dialectical variations.

Supporting Global Projects

The organization recently distributed $3.2 million in grants across four distinct regional projects. In Kenya, the Masakhane organization is working to build datasets for more than 50 African languages, applying these models to health, education, and farming. Meanwhile, the Institute for Worldmaking in Lebanon is digitizing Arab history into machine-readable databases to ensure community oversight rather than corporate control.

In the Brazilian Amazon, the group Portal sem Porteiras is deploying offline AI tools designed to keep sensitive data strictly within indigenous territories. Additionally, the African Internet Rights Alliance is developing auditing tools that ensure existing AI systems are held accountable across the continent. Bdeir stresses that these projects do not prioritize massive scale, which she labels a Big Tech preoccupation. Instead, the focus remains on utility, such as an indigenous elder utilizing a tool to preserve ecological knowledge.

Future Collaboration and Open Development

Transparency remains central to the nonprofit's strategy. At the AI for Good Global Summit in Geneva, Current AI unveiled AlphaChat. This open-source chatbot reached completion in seven weeks through the collaborative efforts of a coalition including Hugging Face, Mozilla, and the MIT Media Lab. Each member provided distinct components, from the base language model to safety architecture.

Beyond individual projects, Current AI has formalized a partnership with Tokyo-based startup Sakana AI. Together, they intend to build a shared open-source stack that prioritizes the Japanese language while remaining adaptable for marginalized communities across the Global South. Bdeir views these partnerships as essential to building a system that replicates the early architecture of the World Wide Web, where shared improvements allow the entire network to evolve without locking participants into a proprietary ecosystem. She maintains that the responsibility for data and knowledge should rest with communities, not with centralized firms in Silicon Valley.