Current AI, a nonprofit formed to create public alternatives to commercial AI systems, has rapidly deployed prototypes, grants and open-source tools aimed at keeping language, culture and data control out of private hands. Founded by Martin Tisne in February 2025 and led by CEO Ayah Bdeir, the organization is assembling funding, partnerships and pilot projects that aim to make AI accessible across languages, regions and offline contexts.
A public alternative to private AI
Current AI’s core argument is straightforward: the largest AI systems today are owned and operated by private companies, and if AI will reshape everyday life it should also include a public option. The nonprofit describes its work as a public-private partnership that pools support from governments, philanthropies and industry players to fund public-interest technology.
The French government provided the initial seed funding of $100 million, and the organization lists further commitments from the Ford Foundation, the MacArthur Foundation, DeepMind and Salesforce, which together bring total pledged support to $400 million. Current AI stresses that these contributors are funders rather than investors, with the goal of keeping outcomes open and broadly available.
Pilots and grants aimed at linguistic and cultural inclusion
One of Current AI’s most concrete early efforts came from a collaboration at the India AI Summit with Bhashini, the Indian government’s AI language division. The partnership produced Suno Sutra — a compact, offline device that runs AI in 22 Indian languages and is open-sourced so developer communities can extend it. The device is intended for scenarios where internet access is limited and where users do not speak English, a design choice framed around ensuring local languages are usable by AI tools.
In its first cohort grant round Current AI allocated $3.2 million across four organizations working in Kenya, Lebanon and the Brazilian Amazon. The Kenyan project with Masakhane focuses on building datasets across more than 50 African languages for domains such as health, farming and education. Lebanon’s Institute for Worldmaking is digitizing Arab cultural history and current practice into machine-readable formats under community control. In the Amazon, Portal sem Porteiras is creating offline AI tools in partnership with Indigenous communities to keep data within local territories. The African Internet Rights Alliance is developing audit tools designed to increase accountability for AI systems across the continent.
Data ownership and community consent
Current AI places data ownership and consent at the center of its approach. Rather than funneling language and cultural data into centralized corporate models, the nonprofit emphasizes storing models and datasets locally and involving community experts before building systems. Consent protocols are designed to give communities the power to halt or shape projects, and grantees are explicitly asked to make these questions part of their work.
The organization argues this model rejects the default where complexity or scale becomes justification for handing decisions to governments or Silicon Valley firms. Instead, Current AI is betting that smaller, community-led deployments can preserve cultural knowledge and prevent extraction of language data without consent.
Assembling an open AI stack
Current AI has also moved quickly on tooling. In Geneva it launched AlphaChat, an open-source chatbot built in roughly seven weeks by a coalition of ten organizations including Hugging Face, Mozilla and MIT Media Lab. Contributors supplied pieces of the stack such as language models, safety tooling and compute resources to assemble a usable, public chatbot.
The nonprofit also announced a collaboration with Sakana AI, a Tokyo-based startup focused on what it terms Sovereign AI. The two groups plan to develop a shared open-source stack intended to support Japanese language and cultural contexts, while also serving communities in the Global South that are often overlooked by dominant AI providers.
Scope, scale and the road ahead
Current AI’s leaders reject the idea that only large budgets equate to meaningful impact. The $3.2 million grant round and early prototypes are framed as strategic seeding rather than comprehensive solutions. The nonprofit highlights scenarios where a modestly funded tool can have outsized effects — for example, enabling elders in an Indigenous community to preserve ecological knowledge in their own language with tools developed elsewhere.
Challenges remain. Grantees have not yet solved every question around governance, ownership or long-term sustainability. Current AI says that is precisely the point: projects should build governance questions into their design rather than retrofitting protections after systems are created.
By combining open-source tooling, community-driven datasets and partnerships with governments and foundations, Current AI is attempting to carve out a public infrastructure layer for AI similar in spirit to the early World Wide Web — open, free to use and governed by communities rather than commercial priorities. How durable that infrastructure becomes will depend on sustained funding, policy support and the ability of local projects to scale principles of consent and cultural stewardship into replicable practices.
Source: TechCrunch AI