India Needs Localised, Ethical AI: PM

India’s AI Model: Localised, Ethical, and Ambitious, but Is It Grounded Enough?

On January 10, 2026, Prime Minister Narendra Modi made a compelling call for an ethical and localised AI ecosystem. In a roundtable held ahead of February’s AI Impact Summit, he urged Indian startups and innovators to focus on regional languages, indigenous content, and data privacy. With India now home to over 1.8 lakh startups and securing the third position globally on Stanford University’s 2025 Global AI Vibrancy Tool, this demand is timely. But the roadmap is far from straightforward.

This announcement coincides with exponential growth in India’s tech sector, projected to surpass $280 billion in annual revenues this year. More than 6 million workers fuel this ecosystem, 89% of new startups integrating AI into their solutions. Yet, despite these impressive figures, beneath the enthusiasm lie structural risks: uneven readiness, dependency on foreign algorithms, and stark implementation gaps between ambition and reality.

The Institutional Architecture: Crossing ₹10,000 Crore Is Only Step One

India’s formal AI framework rests on the IndiaAI Mission, launched in 2024 with a budget of ₹10,300 crore across five years. This flagship programme aims to establish a high-end common computing facility with 18,693 GPUs, critical for scaling AI solutions. Running parallel are supplementary initiatives: BharatGen for government-funded multimodal AI, the Sarvam-1 AI model, and Hanooman’s Everest 1.0 targeting multilingual capacities. These efforts complement older AI hubs like Bhashini, an AI-powered translation system for Indian languages.

But the architecture, while ambitious, suffers from fragmentation. The Ministry of Electronics and Information Technology (MeitY) oversees national-level initiatives, while states remain unevenly equipped. The 15 AI Centers of Excellence planned under IndiaAI reflect this gap—the rollout is sluggish, with no clarity on how these hubs will interface with tech ecosystems in Tier-2 cities.

Deconstructing the “Ethical” AI Narrative

What does ethical AI truly demand? The PM’s emphasis on transparency and fairness brings much-needed focus, but principles must navigate regulatory inertia. The Digital Personal Data Protection Act, 2023, for instance, provides protections for sensitive data. Yet, experts flag an implementation vacuum. Data anonymization protocols are inconsistently enforced, and smaller firms lack robust privacy tools, leaving them vulnerable to cybersecurity threats.

Furthermore, India risks falling into the trap of "digital colonization" if it remains dependent on foreign AI firms for foundational technologies. While BharatGen and Sarvam-1 signal ambition, they lag far behind global competitors like OpenAI and Google DeepMind in scalability and computational efficiency. Ironically, PM Modi’s call for indigenous development faces the reality that several startups rely on GPT-series APIs developed abroad for localized Indian use cases.

The bias problem worsens this challenge. India lacks enforceable standards for algorithm audits—crucial, given that AI trained on skewed datasets risks amplifying social and linguistic inequities. A 2025 Indian Express study revealed that 70% of AI-based hiring platforms rejected candidates from small towns because of regional accents—a glaring inequity grounded in inadequate training data.

Global Context: China’s Tightrope Act vs India’s Aspirational Looseness

China’s AI governance is instructive. Beijing’s draft Generative AI Regulation (2023) mandates algorithm audits before public deployment and requires datasets to adhere to “core socialist values.” While India’s approach—emphasizing transparency and entrepreneurship—is less authoritarian, it lacks China’s clarity on accountability. For example, if an India-developed AI misclassifies loan applicants or spreads misinformation, no Indian framework directly addresses operator accountability. Compare this to China, where punitive measures apply to developers violating compliance norms.

Nonetheless, India can learn not to over-centralize its regulatory power. The Chinese model risks stifling innovation, whereas India's startup-heavy ecosystem thrives on decentralized growth. What is needed is a well-balanced mechanism that integrates accountability with room for innovation.

Structural Tensions: Ambitions vs Constraints

Three core tensions threaten India’s AI ecosystem:

• Centre-State Divide: State-level variations in infrastructure are stark. Karnataka leads AI adoption, while in Bihar, less than 12% of startups report using AI-driven technologies.
• Underwhelming Budget: ₹10,300 crore over five years may seem significant, but spread across computing infrastructure, innovation grants, and regulatory capacity, the funding appears diluted. Compare this to the US, which allocated over $28 billion for AI research under the FY25 National AI Initiative.
• Export vs Domestic Needs: Global Capability Centres (GCCs) in Bengaluru serve global MNCs but often overlook domestic social AI innovations. This divergence complicates alignment with the PM’s vision for localised AI.

What Success Could Look Like

To ensure its AI ecosystem thrives, India must focus on measurable metrics:

First, track startup survival rates beyond early funding rounds, especially in Tier-2 cities where infrastructure gaps persist. Second, expand audit mechanisms for fairness—reduce algorithmic bias in sectors like healthcare and logistics. Third, tie AI models to real governance outcomes: improved e-Courts efficiency in managing case backlogs could serve as a promising benchmark.

Importantly, long-term metrics must prioritize localized AI adoption across agriculture, rural health, and education—not just industrial growth.

Practice Questions for UPSC

Prelims Practice Questions

Frequently Asked Questions