What’s next for India’s AI ambitions
Author: Aahil Sheikh
Disclaimer: The views expressed in this article are those of the authors and do not necessarily reflect the views of their affiliated institutions. The authors write in their personal capacity.
Much has changed about the AI Summit franchise since its inception in the United Kingdom (UK) in 2023. Goalposts have shifted somewhere between an invite-only session at Bletchley Park to deliberate on how emerging Artificial Intelligence (AI) systems will be governed safely and responsibly, to a large-scale, expo-style event to demonstrate national competence and invite foreign investment. Priorities have shifted from safety to innovation and economic growth; the argument goes that excessive concerns over safety may hamper the potential of AI to deliver “unprecedented” economic prosperity that benefits everyone.
This was especially the case for India, which explicitly organised its AI Summit around the notion of “impact”. Among the many outcomes of the AI Summit was the New Delhi Declaration, which was endorsed by 92 countries and international organisations. The Declaration places India in a steering position on conversations on “AI for All”, which is rooted in equity, access, and global cooperation. However, the most immediate, more tangible outcome is perhaps the billions of dollars worth of investments from domestic and global players alike, set to build out AI infrastructure in the country in the form of data centres.
The past two Summits in Paris and New Delhi sought to position the respective host countries as stable and vibrant centres of innovation so as to invite foreign investment into the domestic ecosystem, while bringing together public and private sector entities to further responsible AI development and deployment. However, what is considered “responsible” is also being re-evaluated in real time. At last year’s AI Action Summit in France, the UK and the United States (US) refused to sign the International AI Action Statement. Since then – amidst tariffs and deepening (transatlantic) rifts – international politics has fundamentally changed, and technology governance has become a fault line in international cooperation on AI.
At this juncture, driving consensus on crucial technologies is essential, and the AI Impact Summit sought to achieve just that. The summit was successful in convincing the UK and the US to sign the New Delhi Declaration. The Declaration is nevertheless lacking in certain aspects, most prominently in the absence of the word “safety”. This may be a deliberate call to cater to the appetite of certain countries that see safety as an impediment to technological development. Regardless, such concerns have not hampered the forging and strengthening of alliances between India and other countries; within the first 2 months of 2026, India and the EU signed a historic Free Trade Agreement (FTA), and India recently joined the US-led Pax Silica Coalition.
Between India’s chairmanship of the Global Partnership on AI and the Impact Summit, India is seeking to influence AI policy development at international fora through strategic engagement and new partnerships. Leading progressive policy development that creates an impact extends beyond financial investment and requires forward-looking action on the systems underlying AI, its impact on society, and fostering capacity-building. The many verticals of digital policy, such as data protection, climate, and literacy, all feed into the puzzle that is AI governance. It is thus worth examining how India can make its AI leadership more resilient by taking a holistic approach to policymaking and charting a blueprint for the Global South.
Data governance
The bedrock of modern technology, especially AI, is data. Designing robust data protection regulatory frameworks will go a long way in creating AI models that do not compromise on privacy, while providing a high-standard of privacy protection to citizens. India’s Digital Personal Data Protection Act of 2023 (DPDPA), as it stands, does not provide a high degree of protection to citizens. This has been noted by civil society organisations as well as other regulators.
In February 2024, the European Investment Bank (EIB) submitted a request to the European Data Protection Supervisor (EDPS) to authorise the transfer of personal data to several non-EU countries, including India. The EDPS, which acts as the EU’s independent data protection authority, rejected this request. The rationale cited was a lack of evidence that the countries in question could demonstrate an “equivalent level of data protection.”
An adequacy status is an important benchmark on the European side, as the EU’s General Data Protection Regulation (GDPR) is largely considered to be the “gold standard” for data protection regulation. The European Commission, in consultation with the European Data Protection Board, thus bestows an adequacy label upon countries deemed to have an equivalent level of data protection. In the aforementioned case of the EIB, Brazil was one of the countries that was denied authorisation for data transfer in 2024, on the grounds of inadequate data protection mechanisms. This situation has since changed – in January 2026, the Commission granted Brazil the adequacy status, meaning that “transfers may take place without the need to obtain any further authorisation.”
In order to meet evolving data protection standards to actively participate in international data flows, India must strengthen its own data governance frameworks. Lack of safeguards could hamper effective partnerships in this domain. Currently, such discussions are steered by Article 9.5 of Chapter 9 of the EU–India FTA, which deals with Digital Trade, and provides details on privacy and protection of personal data. It is, however, surprisingly light on details: it merely acknowledges the right of both parties to decide their own level of data and privacy, stating that “each Party may adopt and maintain the safeguards it deems appropriate to ensure the protection of personal data and privacy, including through the adoption and application of rules for the cross-border transfer of personal data.” Furthermore, safeguards adopted by either side shall merely be communicated to the other.
Previous drafts of the FTA, such as the 2022 EU–India trade agreement proposal on Digital Trade, elaborated upon cross-border data flows in addition to discussing privacy and data protection. The Report of the Twelfth Round of Negotiations on a Free Trade Agreement between the EU and India (July 2025) instead states that – pending technical consultations – the chapter on Digital Trade was “agreed in principle”. New details will emerge with time, but the emphasis must be on reconciling differences in pursuit of promoting digital trade and international data flows, rooted in trust in the partner’s data protection infrastructure.
Currently, the GDPR and the DPDPA overlap in some areas: both take as their starting point the status of privacy as a fundamental right; consent requirements are somewhat similar, and certain rights such as correction, portability, and grievance mechanisms are present across the board. But they also diverge in important ways: key differences include government access to data without much judicial oversight, broader definitions, and a lack of independence for the supervisory authority. Such misalignments can hinder cooperation on the data governance front between India and the EU. By contrast, stricter rules could harmonise standards for governance and open up common avenues to evolve regulatory mechanisms based on shared values.
In the context of AI, both the GDPR and the DPDPA must consider how publicly-accessible data may be used for training AI and how consent can effectively be managed. A recent working paper titled “A New Approach to Data Governance” by the Expert Engagement Group proposes a Sovereign Data Exchange Framework: AI companies can train their models within secure data enclaves, and the raw datasets then remain under national control while developing AI systems. Yet, such frameworks do not operate in isolation and require multilateral cooperation.
For the EU and India, common standards for interoperable data governance mechanisms and data flows must evolve out of enhanced engagement. In order to achieve such common standards, the two sides must acknowledge points of convergence and divergence at the intersection of AI legislation, data governance, and privacy protection. The coming months will bring greater clarity in this regard. This will not be an easy task, and both civil society and government actors must remain cognisant of the challenges ahead.
Climate
Conversations on AI safety are often dominated by a focus on loss-of-control, the risk of AI assisting the development of chemical, biological, radiological, and nuclear (CBRN) weapons, autonomous cybersecurity attacks, and other fears. Similarly, conversations on AI affecting labour, medicine, and welfare provisions also find themselves on the docket of high-level multilateral and domestic policy forums. These are conversations worth having – but they obscure another important dimension.
Less attention is given to the impact of AI on the climate and energy resources. When such conversations do occur, they are always offset by the fact that emerging AI will also help “solve” climate change, and thus any present harm is already accounted for.
Some scepticism is in order. Climate change is the result of diverse factors, and likely not “solvable” by a single catalyst in the form of emergent AI capabilities. If such an assumption is to be believed, concrete evidence will go a long way in guiding policy action.
That evidence base is currently lacking, while a recent report on whether AI can help tackle climate change explores how the current AI hype is “greenwashing” the effects of AI on the environment. The wave of investments in AI is predominantly geared towards building data centres; billions of dollars are flowing into India by domestic players, such as Yotta and Adani, and international hyperscalers such as Google and Microsoft.
India’s attractiveness to these large players is by design. A 20-year tax holiday to incentivise global players using local data centers, a greater push towards data localisation, and relatively cheap electricity prices are valuable propositions for companies seeking to build data centres to train and run their AI models. The benefits for the local population, by contrast, are unclear. The assumption is that data centres create jobs and thus spur economic growth for the local economy, but some have cast doubt on these claims. The stress on the environment is real and urgent, and the ends must justify the means.
There is already some progress in this regard. At the India AI Impact Summit, the United Nations Educational, Scientific and Cultural Organisation (UNESCO) launched the Resilient AI Challenge, which was co-organised by the Governments of France and India. The initiative seeks to advance lightweight AI models that can reduce energy consumption by up to 90%. Engagement at the multilateral level through UNESCO and the Coalition for Sustainable AI will bring diverse experts together and promote a common vision for sustainable AI.
Industry collaboration also needs to be driven beyond the pursuit of profit. Both Mistral AI (French) and Sarvam AI (Indian) are making a case for wider adoption of so-called “small AI”. For example, the use of small AI to assist farmers and small enterprises has been a consistent theme for India’s vision of AI. The New Delhi Declaration highlights the importance of developing energy-efficient AI systems and infrastructure.
The intersection of climate, energy, and AI requires a whole-of-society approach if equitable and sustainable solutions are to be developed. The wider rollout of energy-intensive AI infrastructure is already underway and is affecting communities across India. It is thus vital to discuss the environmental impacts of AI alongside its socio-political and economic effects. Putting AI and climate deliberations in a silo of their own threatens to detach them from the larger public debate on AI’s impact on society, which can further undermine holistic policy responses. In the near future, provided that discussions on AI and the climate evolve beyond nascency, it is entirely plausible that one of the future AI Summits is dubbed the AI Climate Summit.
Digital literacy
Viksit Bharat 2047 (Developed India by 2047) places digital literacy at the centre of India’s development ambitions. But the promise of digital infrastructure could well remain unfulfilled if digital literacy lags behind.
Recognising this reality, India and the EU share the view that digital literacy is a priority: the Pradhan Mantri Gramin Digital Saksharta Abhiyan (PMGDISHA) programme was one of the world’s largest rural digital literacy programmes to expand access to online services and financial inclusion, while the EU’s Digital Decade Policy Programme aims for 80% of adults to have at least basic digital skills by 2030.
The quality of digital literacy initiatives relies on access to tools, but equally crucial is an emphasis on educational institutions, their capacity to provide holistic education, competent teaching professionals, and opportunities for graduates to use their skills in professional contexts. That being said, digital literacy and AI literacy encompass a far wider range of skills than is commonly thought. Knowing how to navigate a socio-political ecosystem, where technology governs various aspects of modern life, requires greater emphasis in modern pedagogy.
Media and Information Literacy (MIL) and AI literacy go hand-in-hand, and are crucial skills to critically engage with the outputs of AI that may be “hallucinated” or deepfakes created with the intention of misleading the public. While India is experimenting with laws to tackle the latter, reducing hallucination requires constant technological improvement of Generative AI (GenAI) systems.
The most nuanced solution relies on understanding the role that regulation, and education and literacy (not just digital literacy) play in keeping users safe, investing resources in both accordingly. Policies and guidelines that call for the labelling of AI-generated content, such as the draft Code of Practice on Marking and Labelling of AI-generated content at the EU-level and India’s amendment to the IT Rules 2021, are noteworthy efforts to enforce a base level of transparency. Still, labels themselves cannot always be relied upon to differentiate between the authentic and the synthetic; a critical ability to navigate the political landscape during times of crisis or high-volume events, such as elections, will be paramount to a well-informed populace. This should form the underlying principle of digital literacy policies.
Digital literacy must extend beyond future-proofing skills to ensure continued competence in an evolving labour market. Moreover, digital literacy and AI literacy should not only seek to boost productivity in academic and professional domains, but also cultivate the ability to look beyond the errors of faulty algorithms and AI-generated media. This skill is especially crucial for government officials and civil servants who must grapple with the deployment of emerging technologies in the course of their public duty.
As such, to empower civil servants with the knowledge of diverse aspects of AI in governance, such as procurement, data governance, and stakeholder management, India launched the Competency Framework for AI Integration in India, which is adapted from UNESCO’s Framework on AI and Digital Competencies for civil servants.
Given the common ground and recent advancements in bilateral co-operation, India and the EU can work together to chart best practices and turn principles into praxis. The EU-India Trade and Technology Council (TTC) is an ideal avenue to achieve just that.
AI governance represents an opportunity for greater transparency. A critical audience can interrogate the outputs of algorithms and demand accountability from a “black box” technology. Digital literacy is an integral aspect of the digital economy – the ability to make the most of the digital infrastructure in a safe, secure, and informed way is the only way that the potential of these nascent services and technologies can be fully realised.
Ways forward
The intersection of climate, data protection, and digital literacy forms a cornerstone of AI governance. An AI Summit, impactful as it may be, merely represents the starting line from which principles will emerge. Data governance rules will ensure a baseline degree of privacy protection for users, while digital literacy will facilitate their engagement with this rapidly-evolving technology. The engine behind this development – data centres, model training – consumes significant levels of natural resources and energy, and will require further scrutiny to ensure their sustainability.
The next AI Summit is scheduled to take place in Switzerland next year. Conversations on how AI will help advance “people, planet, and progress” require a multidisciplinary approach. Only then will the future of AI safety, security, and innovation be inclusive, accountable, and sustainable for the coming generations.
Aahil is a digital policy researcher, and an Editor at Generation EU-India. His experience at the grassroots and multilateral levels across India and France allow him to translate principles into practice. He is interested in a range of digital policy issues, such as the dynamic contours of tech diplomacy, and how internet culture affects offline trends. Aahil holds a Master’s in Public Policy from Sciences Po, Paris, and a Bachelor’s in Political Science from the University of Delhi.
Generation EU-India 