Comparative AI Policy Analysis
Systematic review of national AI strategies, regulatory proposals, and implementation mechanisms across Global South and OECD jurisdictions — identifying gaps, transferable models, and divergence risks.
ResearchAbout Atlas AI Institute
Atlas AI Institute is an independent, non-partisan research organisation advancing evidence-based AI governance and policy for the Global South — anchored in South Asia and committed to public-interest scholarship that reflects the governance realities of low- and middle-income countries.
Mission & Purpose
Why Atlas AI Institute Exists
The global discourse on artificial intelligence governance is disproportionately shaped by institutions in North America and Europe, often producing frameworks that are technically rigorous but contextually misaligned with the institutional capacities, legal traditions, and developmental priorities of the Global South. Atlas AI Institute was established to address this structural imbalance — not by rejecting international standards, but by ensuring that countries across South Asia, Sub-Saharan Africa, Southeast Asia, and the Pacific are active architects of the norms that will govern transformative technology in their own societies.
Our work is grounded in the conviction that durable AI governance requires more than policy transplantation. It demands original research calibrated to local regulatory environments, public consultation processes adapted to varying levels of digital literacy, and institutional capacity built from the inside rather than imposed from without. We produce comparative analyses, regulatory design tools, and capacity-building resources that policy actors can actually use — irrespective of whether their ministries have dedicated AI offices or dedicated AI budgets.
Atlas AI Institute operates as a research institute in the public interest. We do not represent any government, corporation, or political movement. Our findings are produced through transparent methodologies and are subject to documented editorial and ethics standards. Where evidence is contested or incomplete, we say so. This commitment to intellectual honesty is foundational to our institutional identity.
Focus Areas
What We Do
Our research and advisory work spans six interlocking domains, each oriented toward practical policy relevance rather than purely academic output.
AI Governance & Regulatory Design
Advisory support and working documents for governments and regulatory bodies developing AI legal frameworks, algorithmic accountability measures, data governance structures, and public-sector AI protocols.
PolicyGlobal South & LDC Institutional Capacity
Research and tools for Least Developed Countries and emerging-economy governments with limited technical staff, constrained regulatory budgets, and rapidly digitalising public services.
CapacityPolicy Observatories & Benchmarking
Ongoing monitoring of AI governance developments across target regions, including annual benchmarking of national AI readiness, regulatory coherence, and implementation progress.
MonitoringPublic-Interest AI & Digital Sovereignty
Analysis of AI systems in public services — social protection, health, justice, education — with emphasis on algorithmic accountability, civic data rights, and digital infrastructure governance.
CivicCapacity-Building & Research Translation
Practitioner-oriented training resources, policy briefs, and translated materials bridging technical AI research and the working knowledge of civil servants, journalists, and civil society actors.
Education
Methodology
Our Approach
Atlas AI Institute does not produce research for its own sake. Every output is oriented toward a practical question: what can a policymaker, regulator, or civil society actor do differently as a result of reading this?
Our methodology combines quantitative evidence review, qualitative institutional analysis, and structured stakeholder engagement with governments and civil society across target regions. We apply documented source classification standards to distinguish between peer-reviewed findings, grey literature, institutional reports, and expert judgment.
- 01Evidence before prescription. Policy recommendations follow from documented evidence, not from prior commitments to particular governance models or technology positions.
- 02Context as a primary variable. Governance solutions are evaluated within the legal, institutional, and resource realities of the country under analysis — not against idealised standards abstracted from context.
- 03Transparent uncertainty. Where evidence is incomplete or contested, we document this explicitly rather than extrapolating from better-studied contexts.
- 04Stakeholder-grounded analysis. Research outputs are developed in dialogue with practitioners, civil servants, and civil society in the jurisdictions studied.
- 05Open access by default. All primary research publications are freely available without paywalls.
"Effective AI governance in the Global South cannot be built on imported frameworks designed for institutional environments that do not exist in most of the world. It must be built on honest analysis of what institutions actually are, what they can realistically become, and what support they need to get there."
— Atlas AI Institute Research Principles, v1.0
Frameworks We Reference
Analyses are benchmarked against, but not limited to:
- OECD AI Principles (2019, updated 2024)
- UNESCO Recommendation on the Ethics of AI (2021)
- EU AI Act Regulatory Architecture
- NIST AI Risk Management Framework (2023)
- ITU AI for Good / AI Policy Observatory
- AU AI Strategy and African Union frameworks
- G7 Hiroshima Process on AI
Geographic Focus
Centring the Global South
More than 80 percent of the world's population lives outside the OECD. The countries where AI deployment is growing fastest — and where governance capacity is most limited — are overwhelmingly in the Global South. Yet the institutions producing actionable AI governance research remain concentrated in Washington, London, Brussels, and Beijing.
Atlas AI Institute positions the Global South not as a passive recipient of AI governance norms, but as a constituency with distinct interests, institutional realities, and governance innovations that the global policy conversation has largely failed to incorporate.
Bangladesh and South Asia serve as our primary anchor — a region combining high rates of mobile-first digital adoption with significant public-sector digitalisation, a growing AI startup ecosystem, and governance institutions navigating rapid technological change with limited dedicated resources. From this anchor, our scope extends to:
- South Asia — Bangladesh, India, Sri Lanka, Nepal, Pakistan, Maldives
- Southeast Asia — with emphasis on ASEAN member states
- Sub-Saharan Africa — including AU digital policy processes
- Pacific Small Island Developing States
- Latin America & Caribbean — comparative governance analyses
Why Bangladesh as Anchor?
Bangladesh exemplifies the governance challenge facing digitally ambitious LDCs: a National AI Strategy under development, a rapidly growing digital economy, significant fintech and agri-tech AI deployment, and public institutions building AI oversight capacity in real time. Research grounded in this context produces insights applicable to dozens of comparable countries.
80%+Global population in the Global South
5+Priority regions in current research scope
40+Countries without a published national AI strategy (2025)
OpenAll primary research freely available
Figures sourced from UN, ITU, and OECD AI Policy Observatory. Subject to revision.
Institutional Governance
Governance & Structure
Atlas AI Institute is designed to operate under an independent governance structure that ensures research freedom, editorial independence, and accountability to its stated public-interest mission — rather than to any single government, funder, or institutional sponsor.
Research Leadership
Atlas AI Institute's research outputs are led by its founding director and a developing team of policy researchers with backgrounds in AI governance, public law, and development policy. Staff and researcher profiles will be published as the team grows and roles are confirmed.
Advisory Structure
An external advisory body comprising independent AI policy scholars, legal practitioners, and civil society leaders from Global South jurisdictions is in formation. Advisory membership will be published only when formal commitments are documented and publicly verifiable.
Legal Registration
Details of Atlas AI Institute's formal legal registration, jurisdiction of incorporation, and organisational type will be published once registration is confirmed and publicly verifiable, as part of our full institutional transparency commitment.
Research Integrity
Research Standards & Ethics
Atlas AI Institute applies documented editorial and methodological standards to all research it produces or publishes. These standards govern source classification, author conflicts of interest, factual accuracy review, and the treatment of contested or uncertain evidence.
Source Classification
All citations are classified by evidence type: peer-reviewed research, institutional grey literature, government documents, news reports, or expert judgment. Claims reliant on lower-confidence sources are qualified accordingly. Unverifiable statistics are not published without source attribution.
Conflicts of Interest
Authors and contributors are required to disclose financial relationships, institutional affiliations, and advisory roles that could constitute a conflict of interest relative to the subject matter of a publication. Disclosed conflicts are noted in the relevant publication.
Corrections Policy
Factual errors identified after publication are corrected promptly, labelled clearly, and dated. We do not silently edit published research. Substantive corrections that change a finding or recommendation are accompanied by an explanatory note.
Editorial Independence
No funder, partner, or advisory member has editorial authority over Atlas AI Institute research outputs. Funded projects are published only if they meet our internal standards; funders receive no right of veto or pre-publication review.
Transparency
Transparency & Accountability
We hold ourselves to the same transparency standards we apply to the institutions we study.
Funding Disclosure
Atlas AI Institute is committed to publishing a formal funding disclosure statement identifying all sources of institutional income. We will not accept funding from sources that require editorial influence over research outputs, and we will publish any conditions attached to grants received. Full disclosure will be published as operational funding arrangements are confirmed.
Corrections & Accountability
All published research is subject to our documented corrections policy. Readers who identify factual inaccuracies, outdated citations, or errors in analysis are encouraged to contact us. We aim to respond within ten working days and publish confirmed corrections within a further ten days. A public log of corrections is maintained and linked from individual publications.
Read our Transparency Statement
How to Engage
Work With Us and Our Research
Atlas AI Institute's research is produced in the public interest and freely available. There are several ways to engage with our work.
Read Our Research
Access our published policy analyses, comparative governance reports, and working papers — freely available, no registration required.
Browse ResearchJoin the Coalition
Connect with researchers, civil society actors, and policymakers working on AI governance across the Global South through our practitioner network.
Join the CoalitionContact Us
For research collaborations, media enquiries, corrections requests, or general questions about Atlas AI Institute's work and mission.
Get in Touch
Frequently Asked Questions
About Atlas AI Institute — FAQ
What is Atlas AI Institute and who does it represent?
Atlas AI Institute is an independent, non-partisan AI governance and policy research organisation. It does not represent any government, political party, corporation, or individual funder. The institute operates in the public interest, with a specific focus on producing research that serves the AI governance needs of the Global South — anchored in Bangladesh and South Asia.
It is not an advocacy organisation in the sense of lobbying for specific legislative outcomes; rather, it produces evidence and analysis that informs policymakers across the political spectrum.
How is Atlas AI Institute funded?
Atlas AI Institute is committed to publishing full funding disclosure once its operational funding arrangements are confirmed and verifiable. Our policy is to accept grants and project funding only where no editorial conditions are attached — meaning funders do not have the right to review, suppress, or modify research findings prior to publication.
Any funding accepted will be disclosed publicly, including the funder's identity, the amount received, and the scope of the funded work. We will not accept funding from sources whose interests create irresolvable conflicts with our research mission.
Is Atlas AI Institute affiliated with any government or university?
No. Atlas AI Institute is an independent research institute and is not a division, department, or affiliated unit of any government ministry, regulatory body, or academic institution. Researchers affiliated with the institute may hold or have held positions at academic or government institutions; any such affiliations are disclosed on a per-author basis in individual publications.
Independence from government and corporate affiliation is a deliberate structural feature that allows us to publish analysis critical of any government's AI governance positions — including the Government of Bangladesh — without institutional constraint.
How can I cite or reuse Atlas AI Institute research?
All primary research publications are made freely available under an open access model. Specific licensing terms are noted on individual publications; most are released under a Creative Commons Attribution (CC BY) licence, which permits reproduction and adaptation with appropriate attribution.
For citation guidance, refer to the suggested citation format on each publication's page. If you wish to use our research in a commercial context, for training AI systems, or in ways not covered by the standard licence, please contact us before proceeding.