BharatGen AI to Cover All 22 Scheduled Indian Languages by 2026, Aiming to Bridge Digital Language Gaps

India is working on a big effort to make artificial intelligence (AI) more accessible by supporting all 22 official languages listed in the Eighth Schedule of the Constitution. The Minister of State for Science and Technology and Earth Sciences, said that the goal is to do this by June 2026. The objective is to bridge the digital gap and make AI tools available to everyone in the country, no matter what language they speak.

This initiative is part of BharatGen AI, India's first government-backed multimodal AI model that can understand and work in native languages and cultural contexts. The National Mission on Interdisciplinary Cyber-Physical Systems (NM-ICPS) from the Department of Science and Technology operates this initiative.BharatGen is being made to work with text, speech, and visuals in many languages. It will be useful in many fields, including government, farming, healthcare, education, and more.

Presently, BharatGen works with nine languages: Hindi, Marathi, Tamil, Malayalam, Bengali, Punjabi, Gujarati, Telugu, and Kannada.  It desires to incorporate Assamese, Maithili, Nepali, Odia, Sanskrit, Sindhi, and others by December 2025. By the middle of 2026, they will have covered all 22 languages. The rollout will come in stages, with the first stage focused on making the current languages better and the second stage adding the other ones.

The IIT Bombay Technology Innovation Hub (TIH) for IoT & IoE is leading the project, with help from IIT Madras, IIT Kanpur, IIIT Hyderabad, IIT Mandi, and IIM Indore. Every institution has a separate role, including making voice recognition better, coming up with tokenization strategies, training the model, and checking how accurate it is.

It's not only about technology at BharatGen, it's also about helping people. Early tests have already been employed in farming, the military, and services for citizens. An example is combining BharatGen with CPGRAMS, the government's system for handling complaints, so that complaints may be handled in more than one language and the process is easier for people.

But building an AI model that is this big and open to everyone is hard. Ganesh Ramakrishnan, a professor at IIT Bombay, says that one of the biggest problems is that there isn't enough data for many Indian languages. Every language has its own writing system, varieties, and rules of grammar. To fix this, the team uses OCR to get information from public books, UDAAN to translate it, and collaborates with government libraries and archives to get materials.

Experts say that BharatGen is a huge step that the technology is made in India, uses Indian data, and is focused on the requirements of Indian residents instead of relying on systems from other countries that are focused on English. If it works, BharatGen might transform how millions of Indians use AI, making it truly representative of the country's linguistic diversity and open to everyone.  By the middle of 2026, India may have AI tools available in all of its official languages. This would make technology easier to get than ever before.

This article is based on information from Observe Now