Responsible AI Impact Assessment Toolkit
Practical worksheets for public agencies to identify affected groups, evaluate risks, and document accountability decisions.
Responsible AI Observatory · South and Southeast Asia
Independent research, regional data, and practical knowledge for technology that serves people and the public interest.
South Asia · Southeast Asia
Recently published
Practical worksheets for public agencies to identify affected groups, evaluate risks, and document accountability decisions.
Why responsible AI depends as much on administrative capacity as it does on technical standards.
Independent audits of AI systems only work if regulators have the technical capacity, legal standing, and political will to act on findings.
Featured research
This brief surveys participation mechanisms in five countries and identifies conditions for meaningful (rather than performative) inclusion.
Read the report"Responsible AI requires institutions that can translate principles into accountable public practice." Regional baseline finding
Interactive dashboard
Explore evidence foundations across eight countries. Scores summarize documented policy and institutional signals; they are not rankings of social progress.
Last updated28 July 2026
| Country | Policy | Capacity | Data | Inclusion | Overall |
|---|---|---|---|---|---|
| Bangladesh | 62 | 48 | 54 | 57 | 55 |
| India | 74 | 69 | 63 | 55 | 65 |
| Sri Lanka | 66 | 61 | 58 | 64 | 62 |
| Thailand | 72 | 67 | 65 | 62 | 67 |
| Malaysia | 81 | 76 | 70 | 65 | 73 |
| Indonesia | 69 | 64 | 62 | 59 | 64 |
| Philippines | 70 | 63 | 60 | 72 | 66 |
| Vietnam | 77 | 70 | 68 | 60 | 69 |
Regional AI Policy Inventory 2026 and country expert validation.
Four equally weighted normalized indicators. Missing evidence is not scored as zero.
Research framework
Regional scope
Country portals connect national context with comparable regional evidence.
Practice and learning
A multilingual decision-support service combines local forecasts and agronomic knowledge while retaining human extension support.
A clinical support pilot evaluated how risk models can assist—not replace—community health workers.
Open evidence
A structured inventory of national AI strategies, bills, standards, and public-sector guidance across 15 countries.
Documented public-sector AI applications with structured fields for benefits, risks, and safeguards.
Regional expertise
Research Director
Data governance specialist focused on public-interest data infrastructure.
Upcoming event
A fully funded 12-month fellowship for emerging researchers advancing responsible AI in South and Southeast Asia.
Event detailsPartners and funders