Current AI builds open AI infrastructure with $400M, offline device for 22 Indian languages

Current AI, a nonprofit founded in February 2025 by Martin Tisne, is racing to create an open, public AI infrastructure modeled after the early World Wide Web. Led by CEO Ayah Bdeir (formerly Mozilla’s AI head and littleBits founder), the organization aims to counter the dominance of private AI systems from OpenAI, Google, and Anthropic. At the India AI Summit in February, Current AI partnered with the Indian government’s Bhashini to launch Suno Sutra — a pocket-sized, offline device that runs AI in 22 Indian languages without any internet connection. The device is fully open-sourced, allowing developer communities to build upon it. The nonprofit operates as a public-private partnership with $400 million in committed funding from the French government ($100M seed), plus the Ford Foundation, MacArthur Foundation, DeepMind, and Salesforce. In its first grant round last month, Current AI deployed $3.2 million to four organizations: Masakhane in Kenya (building AI datasets in 50+ African languages for health, farming, and education), the Institute for Worldmaking in Lebanon (digitizing Arab cultural history with community-controlled data), Portal sem Porteiras in Brazil (offline AI tools for Indigenous Amazon communities), and the African Internet Rights Alliance (AI audit tools across Africa).

Current AI’s mission is rooted in the fact that half the world’s spoken languages face extinction, and current AI models — driven by English — leave most cultures behind. Bdeir explicitly distinguishes Current AI’s approach from Big Tech’s multilingual push. “Big tech builds multilingual models to expand their market, regardless of consent or context,” she told TechCrunch, noting that Indigenous languages often get scraped from missionary Bible translations without community permission. Current AI prioritizes local data ownership: models and data stay within communities, consent protocols are built into the pipeline, and communities can halt processes at any point. While none of the grantees have fully solved data sovereignty yet, Bdeir emphasizes that embedding the question into their work is the point. The broader vision is an AI ecosystem where improvements benefit everyone, no one is locked out, and communities retain control over their cultural knowledge.