A farmer in rural India takes a photo of a failing crop and hopes to identify it. She does not use English, and current AI systems offer little support in her language. Current AI, a nonprofit, sees this gap as a key issue to address.

The organization aims to deliver a public option for AI that remains open and free, similar to the early days of the web. Chief executive Ayah Bdeir explained to TechCrunch that if AI will reshape daily life for everyone, a public version must exist, available like the World Wide Web to all at no cost.

Who is behind it

Current AI debuted at France's AI Action Summit in early 2025. The French government provided initial funding of 100 million euros, with additional support from the Ford and MacArthur foundations, DeepMind, and Salesforce. Commitments have reached more than $400m, and the group targets $2.5bn over five years.

The supporters seek no financial returns. Bdeir noted they function as funders rather than investors. She previously led AI strategy at Mozilla and founded the education hardware company littleBits.

The effort draws on European interest in public options versus American platforms. It aligns with state initiatives for public AI infrastructure yet operates as a global effort rather than a national one.

Starting with language

The first priority centers on global languages. Half of languages in use today face extinction risks, while English leads in major models. Bdeir argues this excludes most world cultures from AI benefits.

She criticizes how large firms create multilingual systems mainly to grow markets without regard for consent or local context. In some Indigenous languages, missionary Bible translations serve as training data without community approval.

Current AI builds alongside communities and keeps data local. In India, it partnered with a government language unit on Suno Sutra, a compact offline device supporting AI in 22 Indian languages without internet access. The project is open-sourced for further development.

Money on the table

Funding has begun at a measured pace. Last month the group awarded $3.2m in initial grants to four groups in Kenya, Lebanon, and the Brazilian Amazon.

Projects include datasets covering more than 50 African languages, offline tools that retain Indigenous Amazon data within the region, and audit tools for AI accountability. Bdeir acknowledges no grantee has resolved data ownership fully, yet she views raising the issue itself as progress.

On whether $3.2m can create impact, she rejects scale as the sole metric, calling it a Big Tech approach.

Assembling a stack

The larger goal involves a complete open stack from hardware to applications. This month in Geneva, Current introduced Alpha Chat, an open-source chatbot built in seven weeks by ten groups including Hugging Face, Mozilla, and MIT Media Lab. Each contributed elements such as a model, safety tools, or computing resources.

It also formed a partnership with Japan's Sakana AI to develop a shared open stack for Japanese and Global South languages overlooked by major models. The question remains whether a nonprofit can match companies offering free models for market reach. Current AI believes a public option merits development regardless of its initial size.