Responsible AI Trust
Rebalancing AI governance
India and the BRICS call for a UN-led, equity-centred model.
Sonia Abraham
Lehar Gupta
Dr. Giovanna Jaramillo-Gutierrez
Foreword from the founder
At Responsible AI Trust, our purpose is simple yet vital: to strengthen global confidence in intelligent systems.
Across industries and borders, AI is redefining how decisions are made, risks are managed, and accountability is shared. With this transformation comes a collective responsibility to ensure that technology remains aligned with human values, transparency, and fairness.
Each brief we publish represents collaborative work from researchers, advisors, and practitioners who believe that trust must be earned through clarity, governance, and evidence. Together, we translate complex regulation into actionable insight, helping leaders navigate uncertainty with structure and foresight.
Responsible AI is not a trend. It is the foundation of sustainable innovation. As systems grow more capable, our frameworks for oversight must grow equally intelligent, adaptive, and globally connected.
Thank you for being part of this effort to turn principles into practice, and ideas into accountability.
Lehar GuptaFounder & CEO, Responsible AI TrustLehar@ResponsibleAITrust.com
About this brief
Rebalancing AI governance
This brief examines the 2025 BRICS Rio Declaration on AI Governance and its implications for global AI regulation. It compares how BRICS, with India in a central role, is advancing an equity-centred, UN-led model alongside existing EU and US approaches, and where those models align, diverge, and compete.
Leader’s dashboard
Global AI governance is now defined by three competing models.
BRICS challenges Global North rule-setting by prioritising development, inclusion, and sovereign choice, and by rejecting fragmented regional governance. India’s credibility, scale, and technical capacity position it as the primary convergence actor across these models.
The 2025 BRICS Rio Declaration is the first time a Global South coalition has advanced a UN-led, equity-centred framework to counter regulatory fragmentation, rebalance digital power, and widen access to AI benefits.
Maturity model: BRICS readiness
Key takeaways for leaders
- AI governance is now geopolitical infrastructure, not policy theory.
- Equity and sovereignty are becoming global governance demands.
- Fragmentation increases risk, cost, and exclusion.
- India is the decisive bridge actor in global AI convergence.
- UN-led coordination is the scalable path to legitimacy.
Support UN-centred mechanisms, fund Global South capacity, align sovereignty with cooperation, and push for interoperable safety, evaluation, and reporting tools before governance hardens into rival blocs.
Lehar Gupta, Founder & Research Director, Responsible AI Trust. Presented and led by the co-author of this brief.
Executive takeaways
A turning point in who writes the rules
The 2025 BRICS Rio Declaration marks a turning point in global AI governance. For the first time, the Global South has collectively called for a UN-led, equity-centred framework that balances innovation, digital sovereignty, and inclusion. That vision stands beside the EU’s values-driven regulatory model and the US market-driven innovation model.
India sits at the centre of the shift. It combines advanced AI capacity, diplomatic credibility, and leadership in Global South initiatives. Whether AI governance fragments or converges into an inclusive system will depend, in large part, on India’s role as a bridge.
Introduction
Artificial intelligence is increasingly recognised as a transformative technology with economic, social, and geopolitical consequences. Estimates suggest AI could contribute $15.7 trillion to the global economy by 2030. Governance has become one of the most pressing global policy challenges: frameworks have to balance innovation, equity, security, and rights. Regulation can foster innovation when it creates a stable environment that attracts investment, stimulates competition, and builds public trust.
AI is also emerging in a period of geopolitical rivalry, including export controls on semiconductors. Countries, especially in the Global South, are being asked to navigate competing blocs. The EU AI Act, South Korea’s AI Basic Act, the Brazilian AI Bill, and the United Kingdom’s proposed AI Bill all show how widely governments are now attending to the problem.
The EU AI Act (2024) is the first comprehensive regional framework, focused on fundamental rights and risk-based oversight. The United States AI Action Plan (2025) prioritises innovation, security, and voluntary standards. Until the Rio Declaration, that debate was dominated by the Global North. The Council of Europe’s 2024 Framework Convention on Artificial Intelligence, the first binding international treaty on AI, requires states to uphold human rights, democracy, and the rule of law. Its influence depends on whether countries outside Europe, particularly from the Global South, choose to join and shape implementation.
Against that backdrop, the BRICS Rio Declaration on AI Governance (2025) is the first coordinated Global South-led call for a UN-centred, equity-driven model. It argues for avoiding regulatory fragmentation, protecting digital sovereignty, and ensuring that countries can benefit from AI regardless of development stage.
Key analytical angles
- Principles and priorities. The Rio Declaration articulates an equity and sovereignty-driven vision, positions the UN as the central platform, and rejects fragmented regional frameworks.
- EU comparison. Unlike the EU’s human-centric, compliance-heavy model, BRICS prioritises inclusion, development, and flexibility across national contexts.
- US comparison. Unlike the US decentralised, innovation-first strategy, BRICS emphasises multilateral cooperation, fair competition, development equity, and safety.
- India’s role. Domestic capacity-building, responsible AI institutions, and international leadership let India anchor BRICS priorities while remaining credible with the EU and the US.
Background and context
The BRICS Rio Declaration
BRICS began as Brazil, Russia, India, China, and South Africa. Egypt, Ethiopia, Iran, and the United Arab Emirates joined in 2024, and Indonesia became the tenth member in January 2025. Leaders, particularly Russia, have framed the enlargement as the rise of a global majority and a step toward a post-Western order.
The bloc represents roughly 45% of the world’s population, more than 35% of global GDP in purchasing-power terms, and around 30% of the world’s oil. It has also built institutions, including the Contingent Reserve Arrangement and the New Development Bank, and a web of cooperation across energy, health, climate, and technology.
BRICS treats AI as a catalyst for growth, technological progress, and inclusive societies. That is why members established an AI Study Group in August 2023 to manage risks while advancing shared priorities. The 2025 Rio Declaration then called for UN-led multilateralism as the foundation of AI governance, and explicitly rejected fragmented or exclusionary regional frameworks.
The principles are clear. The operational mechanisms are not. Guidance on safety, ethics, and fairness remains largely principles-based, without legally binding obligations. The EU categorises systems by risk and enforces strict measures for high-risk applications. BRICS applies broad principles across all AI systems. That flexibility fits a diverse membership. It also leaves a gap in enforcement tools, monitoring, and capacity-building.
To turn principles into impact, BRICS needs measurable commitments: capacity-building targets for skills, compute, and data governance; safeguards for high-risk uses; interoperable evaluation tools; standardised safety baselines; shared evaluation suites; and transparent progress dashboards.
The vision sits close to the United Nations call for a global AI governance framework built on inclusivity, capacity-building, and multistakeholder oversight. The UN High-Level Advisory Body (2024) proposed mechanisms such as a global scientific panel on AI risks and international repositories for safety practices. The Rio Declaration is more than a reaction to EU or US models. It is an argument that equity, sovereignty, and development should be global norms.
Comparative analysis
EU rights, US innovation, and the enforcement gap
The EU AI Act
The EU AI Act is one of the first comprehensive legal frameworks for artificial intelligence. It classifies systems as unacceptable, high, limited, or minimal risk. Manipulative techniques and social scoring are prohibited. High-risk systems carry lifecycle obligations. Limited-risk systems must disclose that a person is interacting with AI. Minimal-risk applications stay largely unregulated. The structure reflects a precautionary, rights-based approach.
The Act also tries to make Europe a leader in trustworthy AI. High-stakes uses in healthcare, transport, and law enforcement are expected to pass risk management, testing, and human oversight. General-purpose AI models sit inside the scope, with transparency, copyright, and risk-mitigation duties. The AI Office and AI Board are meant to coordinate enforcement across member states.
Implementation has already exposed limits. Key provisions take effect on a staggered timeline through 2025 and 2026. Many member states still lack institutional readiness. Smaller companies have raised the cost of documentation, testing, and transparency. Critics argue that a heavily precautionary regime can deter investment. After the Paris AI Action Summit, the US and UK refusal to endorse the final declaration showed how far Europe’s model sits from more flexible, innovation-led approaches. As Marta Cantero Gamito notes, Europe has to consolidate internal capacity before it can project global leadership.
The Act remains a landmark. Its long-term influence depends on whether the EU can close the gap between regulatory ambition and practical enforcement, and whether it stays engaged with partners outside Europe.
The United States AI strategy
America’s AI Action Plan aims to support private-sector innovation by reducing bureaucratic barriers. It rescinds earlier executive orders seen as onerous and asks agencies to revise rules that could slow development. The federal role is light-touch: targeted measures for specific risks, broad adoption across industries, and a stated commitment to free speech, objective truth, and national security.
The plan also expands access to large-scale compute, supports open-weight and open-source models, encourages AI literacy and workforce retraining, and accelerates AI use inside government. Measures such as the TAKE IT DOWN Act target malicious uses, including deepfakes. The federal government is cast as both an enabler of rapid development and a protector against AI-related threats.
The trade-off is operational detail. Open-weight models can accelerate innovation and also increase misuse and cybersecurity exposure. Unlike the EU AI Act, the US framework does not embed lifecycle security and risk-management requirements. Enforcement and safety remain largely voluntary. Funding gaps, competing federal priorities, and a weaker presence in multilateral institutions can also reduce US influence over international standard-setting.
There is still no comprehensive federal AI law. State measures are filling part of the gap. California’s SB 53 requires major AI companies to disclose safety protocols, assess catastrophic risks, and report critical safety incidents within 15 days, with whistleblower protections. Anthropic has endorsed the bill as consistent with its Responsible Scaling Policies and system cards. The contrast is philosophical: the EU prioritises precaution and enforceable compliance; the US prioritises market-driven innovation and industry-led accountability.
BRICS, the EU, and the United States
The Rio Declaration (2025), the EU AI Act, and the US AI Action Plan answer the same problem with different theories of legitimacy.
National AI strategy
India’s leadership within BRICS
India’s strategy has moved quickly. It tries to use AI for growth while embedding safeguards, governance, and international collaboration. Three pillars hold it together: domestic infrastructure, responsible governance and safety, and international engagement.
Domestic infrastructure
- IndiaAI Mission. A national platform for datasets, affordable compute, and indigenous foundation models aligned with India’s linguistic and societal needs.
- AI for Viksit Bharat (NITI Aayog, 2025). A roadmap toward an estimated USD 500–600 billion GDP contribution by 2035, including 38,000+ GPUs on a federated compute network, regional language models, a consent-based public dataset platform, skilling in Tier 2 and Tier 3 cities, and sectoral use in healthcare, agriculture, education, and mobility.
- Bhashini. More than 350 AI language models have processed over one billion tasks, supporting 50 government departments and 25 state governments.
Responsible AI institutions and safety
- IndiaAI Safety Institute (2025). Sets national standards for safe deployment and connects domestic practice to global safety debates.
- AI Competency Framework (MeitY, 2024). Defines training and oversight for public officials using AI in government.
- AI Governance Guidelines (2025). A multi-stakeholder advisory group under the Principal Scientific Adviser mapped gaps in Indian law and proposed a lifecycle risk approach, from development to diffusion, with whole-of-government coordination.
International engagement
As GPAI lead chair in 2024, India championed applied research aligned with Global South priorities. Its 2025 BRICS chairship has emphasised cooperation, capacity-building, and equitable access to AI resources. Oxford Insights ranked India second globally in AI research publications in 2024, which gives it both technical weight and diplomatic credibility.
India is uniquely placed to bridge Western safety-oriented frameworks and Global South developmental priorities, and to shape a model that holds safety, innovation, and fairness together.
World view
Implications for global leaders
The Declaration is both a chance to correct inequities in AI access and a test of whether emerging powers can coordinate on safety and innovation.
Equity and inclusion. Equitable access could let the Global South help write international standards rather than only adopt them. A UN-led framework, as BRICS proposes, would redistribute power in rule-making.
Digital sovereignty with cooperation. States can keep control of their AI ecosystems while joining collective governance. That is an alternative to reliance on an external regulatory or technology model, including the path set by the EU AI Act.
India’s bridging role. The IndiaAI Mission, the Viksit Bharat roadmap, and the IndiaAI Safety Institute show domestic capacity. GPAI chairmanship and participation in the Bletchley and Seoul declarations show credibility in global safety dialogues. India still has to harmonise ambitions across members with very different maturity, and balance inclusive governance with its own economic objectives.
Fragmentation, capacity, and enforcement. The Declaration has no binding enforcement mechanism and no clear implementation strategy. Overlapping initiatives can fragment governance further. China, India, and Russia have deep AI ecosystems. Ethiopia and Egypt face gaps in infrastructure, workforce, and regulatory readiness. If members prioritise domestic control over shared standards, UN-led governance will stay weak.
Even so, the Declaration can be a starting point for convergence: a signal that a more inclusive conversation, led in part by the Global South, is now on the table.
What this means for you
Final words
The BRICS Rio Declaration marks a structural shift in how global AI governance will be contested and built. It reframes governance away from dominance and toward legitimacy, equity, and participation under a UN-centred model. It does not resolve the governance challenge. It redraws the map.
The decisive test is execution. Principles matter only if they become operational infrastructure: shared metrics, interoperable safeguards, and transparent coordination. India’s role as a bridge between Global South priorities and Global North systems will decide whether convergence or fragmentation prevails.
AI governance is no longer a future debate. It is the infrastructure through which participation, trust, and influence will be determined. Those who act early will help shape the system. Those who wait will adapt to it.
Reflections
Researcher reflections
Working on this brief underscored that AI governance is not only a technical challenge. It is a geopolitical one. Frameworks shape who sets the rules and who benefits from them. The Rio Declaration is an inflection point: a coalition of emerging economies has articulated equity, digital sovereignty, and multilateralism under the UN. That is a redistribution of normative power, not only a policy statement.
The research also revealed an implementation gap. Inclusion and sovereignty will matter only if they become enforceable mechanisms: measurable targets, shared evaluation protocols, and transparent reporting. Without those tools, fragmentation persists and trust erodes.
- Champion UN-led dialogue on minimum safety baselines, interoperable evaluation, and cross-border incident reporting.
- Back India’s role as a mediator between advanced and emerging economies.
- Fund capacity so developing nations can participate in safety, ethics, and infrastructure.
- Embed sovereignty and equity in international standards.
- Coordinate BRICS, the EU, and the US around interoperable safety and accountability, rather than rival blocs.
Advisor reflections
European Commission | EU AI Office. Advisor and reviewer.The Rio Declaration is a welcome rebalancing. By centring the United Nations, equity, and digital sovereignty, BRICS challenges the one-way export of Global North rules while affirming that AI must advance fundamental rights, democratic institutions, and the rule of law. India’s leadership, through digital public infrastructure, “Responsible AI for All,” and multistakeholder forums, shows how rights-centred, open standards and multilingual access can scale across the Global South without lowering standards, if they become operational commitments.
An equity-centred UN track should set measurable capacity-building targets for skills, data governance, and shared compute; safeguards for high-risk uses; protections for creators and personal data; and decision-making seats for low- and middle-income countries. The risk is aspirational drift: inclusive principles that leave standards fragmented and accountability thin.
- Convene at the UN around minimum safety baselines and cross-border incident reporting grounded in fundamental rights.
- Finance public goods: compute credits, multilingual datasets, and open evaluation suites.
- Procure for rights: rights-by-design, bans on abusive surveillance in government AI, and public-interest impact assessments.
- Resource independent oversight, protect whistleblowers, and enable civil-society participation.
- Address model, cloud, and chip bottlenecks that entrench lock-in.
Research director reflections
This brief confirms that AI governance has crossed a structural threshold. The Rio Declaration is the first coordinated signal that global AI governance will not be settled by regional dominance or unilateral standard-setting. It will be negotiated multilaterally under the United Nations. What is most consequential is the reframing of authority. By centring equity, digital sovereignty, and development, BRICS challenges the assumption that governance maturity flows only from the Global North outward.
This shift does not displace existing regulatory leadership or innovation pathways. It introduces a UN-anchored coordination layer: interoperable safety and trust baselines without centralised control. For innovation-led systems, that layer can reduce cross-border friction without sacrificing flexibility.
Principles alone will not rebalance governance. Without shared metrics, interoperable evaluation tools, and transparent progress mechanisms, equity stays aspirational. India is not an ideological counterweight. It is a bridge actor, with the scale, technical capacity, democratic legitimacy, and multilateral leadership to decide whether convergence or fragmentation prevails.
Countries and organisations that invest early in trust, capacity, and interoperability will shape the rules. Those that do not will inherit them.
Read this brief as an early map of where authority, alignment, and legitimacy in AI governance are moving next.
Appendix
Appendix 1. Primary sources
- Alan Turing Institute (2025). AI Governance Around the World: India Country Profile.
- Anthropic. Anthropic is endorsing SB 53.
- BBC (2025). AI Translation Technology Deployed Across Indian Companies and Government.
- Blanchet, A. (2025). Artificial Intelligence Governance in BRICS: Cooperation and Development for Social Inclusion.
- BRICS (2025). Rio Declaration on AI Governance.
- Cantero Gamito, M. (2025). AI Governance and the EU’s Strategic Role in 2025. STG Policy Paper 2025/13.
- Carnegie Endowment for International Peace (2025). BRICS Expansion and the Future of World Order.
- Council of Europe (2024). Framework Convention on Artificial Intelligence.
- Council on Foreign Relations. Opportunities and Risks: Trump’s AI Action Plan.
- European Union. Artificial Intelligence Act.
- GPAI (2024). Summit Outcomes, GPAI New Delhi 2024.
- IndiaAI Governance Subcommittee (2025). AI Governance Guidelines.
- IndiaAI Safety Institute (2025). India’s Strategic Leap Toward Responsible AI.
- Ministry of Electronics and Information Technology (2024). AI Competency Framework. Government of India.
- Ministry of External Affairs, Government of India (2025). BRICS Joint Media Statement.
- NITI Aayog (2021). Responsible AI for All: Approach Document.
- NITI Aayog (2025). AI for Viksit Bharat: The Opportunity for Accelerated Economic Growth.
- Oxford Insights (2024). Government AI Readiness Index 2024.
- Tech Policy Press (2024). The Dangers of Imposing Global North Approaches to AI Governance on the Global South.
- Tony Blair Institute (2024). How Leaders in the Global South Can Devise AI Regulation That Enables Innovation.
- United Nations (2024). Governing AI for Humanity: Report of the High-Level Advisory Body on AI.
- United Nations (2025). Update on High-Level Multi-Stakeholder Panel on AI Governance.
- U.S. Government (2025). America’s AI Action Plan.
Appendix 2. BRICS, EU, and US comparison
The same comparison as the body of the brief, retained here as a standalone reference.
Download the full whitepaper
Public Release 1.0, February 2026.
Disclaimer
This brief has been independently developed by Responsible AI Trust. The analyses and perspectives presented are those of the authors and do not represent the official positions, endorsements, or policies of BRICS member states, the United Nations, or any institutions referenced. This publication is provided for informational purposes only, without warranty, and should not be relied upon as legal or regulatory advice. Responsible AI Trust accepts no liability for decisions or actions taken based on its contents.
All rights reserved. No part of this publication may be reproduced, distributed, or transmitted without prior written permission, except for brief quotations with appropriate attribution. Report concerns to info@responsibleaitrust.com.