Aggregate, don't replicate — our curated collection of the best AI Safety, Governance & Ethics resources, with a Sri Lanka lens.
Signed by leading AI scientists and researchers — a landmark statement from the Center for AI Safety.
safe.ai →Comprehensive overview of AI risk and how to contribute to safety research.
80000hours.org →The main research hub for AI alignment — technical and conceptual posts from leading researchers.
alignmentforum.org →Internationally adopted framework covering transparency, accountability, and human-centred values.
oecd.ai →Academic and policy research on governance frameworks from the Centre for the Governance of AI.
governance.ai →Interactive guide to the EU's landmark AI regulation — the global benchmark for AI governance.
artificialintelligenceact.eu →Research on the social implications of AI — bias, labor, and power in AI systems.
ainowinstitute.org →Fighting algorithmic bias through research, art, and policy advocacy.
ajl.org →Accessible frameworks, weekly AI ethics digest, and the Montreal Declaration.
montrealethics.ai →High-engagement peer chapters and organizations from our global benchmarking analysis — strong references for Sri Lanka's work.
AI developments through the lens of Sri Lanka's context — national policy, societal impact, and regional dynamics.
The Information and Communication Technology Agency of Sri Lanka (ICTA) leads the country's digital transformation. Its National AI Strategy outlines priority sectors and the vision for AI adoption.
Visit ICTA →Sri Lanka's Personal Data Protection Act (2022) is the primary legislation governing data rights. It has direct implications for how AI systems collect, process, and use personal data in Sri Lanka.
The EU AI Act, OECD principles, and emerging international AI standards will shape the AI products and services that reach Sri Lanka — even without local regulation. Understanding these frameworks is strategically important.
Plain-language definitions of key AI Safety, Governance & Ethics concepts — making technical terms accessible for all Sri Lankans.
Ensuring that an AI system's goals and behaviors are perfectly in sync with human values and intentions.
When an AI finds a "shortcut" to achieve a goal in a way that designers didn't intend (e.g., getting a high score without playing the game correctly).
The ability of an AI to perform safely and reliably even in new, unexpected, or adverse environments it wasn't trained for.
Making AI decision-making processes understandable to humans, rather than being a "black box" that gives answers without reasons.
The practice of rigorously testing an AI system by simulating adversarial attacks to find vulnerabilities, biases, or safety risks.
A nation's ability to develop, own, and govern its own AI infrastructure and data according to its own cultural values and security needs.