AI for public good: India’s ecosystem approach to innovation

India is playing an increasingly critical role in developing and deploying artificial intelligence (AI). The country is home to more than 6,500 AI-focused startups, ranking seventh globally in private AI investment, attracting more than $11 billion between 2013 and 2024.

At the same time, India is attempting to address three persistent challenges in the use of AI for public good: limited availability of AI-ready open datasets, weak pathways for identifying high-impact AI use cases and the absence of sandbox environments to test AI-enabled solutions.

These constraints are common across low- and middle-income countries but somewhat amplified in India, given its sociocultural and linguistic diversity.

To address these challenges, India has adopted an ecosystem-based approach focused on building institutional mechanisms to support the development, implementation and scaling of indigenous AI innovation for public good.

This is complemented by broader efforts to strengthen public-private partnerships and improve coordination between federal and state governments.

The approach includes several key elements.

In India, AIKosh and Bhashini are national initiatives designed to facilitate secure access to structured datasets for the private sector, researchers and academia to build AI solutions.

AIKosh functions as a unified platform that aggregates datasets from public and private sector entities across domains, including healthcare, governance and administration, urban planning and mobility.

The platform addresses challenges around fragmented datasets and data provenance by scoring datasets against AI-readiness parameters and providing standardized metadata.

Bhashini, meanwhile, provides access to multilingual datasets and models, with structured model and dataset cards, licensing details and usage guidance. It also supports AI-driven solutions, including language translation application programming interfaces (APIs) and a crowdsourcing initiative for Indian languages.

Importantly, these initiatives function as much more than open data for AI platforms; they are institutions in their own right. AIKosh is managed by the IndiaAI division and Bhashini by the Digital India BHASHINI division, both independent business divisions within the Ministry of Electronics and Information Technology.

This institutional structure provides operational flexibility to drive ecosystem engagement and ensures the long-term sustainability of these initiatives.

At the state level, Telangana’s TGDeX initiative combines features of an open data portal, a sandbox and an expanding use-case library for AI innovation. It aims to foster the development of AI-enabled solutions to address the public challenges.

Together, these central- and state-level efforts suggest a coordinated government-wide effort to improve access to AI-ready datasets for innovation.

Another important component of India’s AI strategy is the establishment of AI sandboxes at both the national and sectoral levels to scale AI-enabled solutions.

AI sandboxes are controlled environments to develop, test and refine AI solutions. India’s model is largely innovation-led and focuses on providing developers with resources, including curated and AI-ready datasets, subsidized compute access, open-source AI models and secure development environments.

Beyond infrastructure, these sandboxes are intended to support responsible experimentation. They allow developers to identify safety and efficacy risks earlier, innovate iteratively and build robust, effective and trustworthy AI systems.

AIKosh and TGDeX both have integrated development environments for researchers, startups and other developers.

At the sectoral level, an expert committee constituted by the Reserve Bank of India has proposed creating an AI innovation sandbox in the financial sector. Similarly, in healthcare, India’s National Health Authority has collaborated with academia to launch the Benchmarking Open Data Platform in Health AI (BODH) to facilitate privacy-preserving AI innovation for patient care.

Developing AI solutions for public good also requires identifying use cases and prototyping AI solutions. In India, this is being done through multistakeholder partnerships to conduct or host AI innovation challenges.

These challenges bring together startups, researchers and developers to work on specific problem statements linked to real-world socioeconomic issues where AI interventions could be impactful.

One challenge, led by the Centre of Excellence for Internet of Things and Artificial Intelligence, focuses on improving maternal and child welfare outcomes by using AI-based analytical tools to monitor infrastructure projects and process large volumes of nutrition-related data, thereby supporting targeted healthcare interventions.

Telangana’s AI innovation challenge focused on six priorities for state departments, including the development of a real-time anomaly-detection AI tool for property registration documents submitted to the revenue department.

These challenges generally provide innovators with curated government datasets, compute capacity and financial support to develop deployable pilot solutions.

Together, these initiatives represent a whole-of-ecosystem approach to AI innovation for public good. Open data platforms improve access to AI-ready datasets. Innovation challenges help identify high-impact public-sector use cases. Sandboxes serve as test beds for developers to ensure solutions align with responsible AI practices.

For the Global South, India’s approach emphasizes a clear lesson: innovation in AI for public good requires far more than just investing in technology. Governments must lead in ecosystem design that effectively connects data, innovators and institutions.

India’s strategic opportunity extends beyond building AI innovations for public good to pioneering a pathway for responsible, at-scale adoption that benefits over a billion people.

India’s next priority should be to chart a roadmap from AI ideation and pilots to wider scale and inclusive adoption. This will require identifying a set of national and state-level deployment priorities in which AI-led interventions can have a sustained socio-economic impact.

The Forum is spotlighting how innovation moves from breakthrough to scale to impact ahead of 'Summer Davos' in China, 22–25 June 2026. Follow the latest.