In August 2025, Nepal's Council of Ministers approved the country's first National Artificial Intelligence Policy, and with it, a set of new institutions most people have never heard of: an AI Regulation Council, a National AI Centre, AI Excellence Centres, and a regulatory sandbox for testing new AI systems. If you're a founder, researcher, or simply curious what this actually means in practice, here's a plain-language breakdown of how each piece fits together.
The 5 institutions built by Nepal's National AI Policy 2025.
What a "Regulatory Sandbox" Means in Practice
A regulatory sandbox is a supervised space where an innovator can test a new product with real users, under a regulator's close watch, without immediately having to meet every rule that applies to a fully licensed, commercially operating business. In exchange for that temporary flexibility, the innovator agrees to strict limits: a capped number of test users, a fixed time window, heavy reporting requirements, and full regulatory authority to shut the test down if something goes wrong.
Nepal already has a working example of exactly this concept, just not in the AI policy itself. Nepal Rastra Bank finalized its own Guidelines on the Regulatory Sandbox for fintech innovation in May 2026, letting banks, payment companies, and licensed fintech startups trial products like digital lending, digital KYC, and embedded finance under temporary regulatory relief for up to six months, with the possibility of a further six-month extension. It's a useful preview of how a Nepali AI sandbox is likely to function once its own detailed rules are published.
How Nepal's AI Regulatory Sandbox Will Work
The National AI Policy 2025 commits the government to "create a Regulatory Sandbox for the safe development and testing of AI systems," assigning primary responsibility to the Ministry of Communications and Information Technology and the National AI Centre, with the AI Regulation Council, e-Governance Board, and relevant sectoral ministries in supporting roles. The policy's own implementation plan lists this as a continuous action item rather than a one-time task with a fixed deadline.
What the policy has not yet published is the fine print — exactly which AI systems will be eligible, how long a testing cohort will run, or what specific regulatory relief participants will get. Based on the structure of Nepal Rastra Bank's already-finalized fintech sandbox, a reasonable expectation is a similar model: a formal application window, a defined testing period with real but limited users, mandatory progress reporting, and a clear "graduation" path once a system proves it's safe and effective enough for full deployment.
Role of AI Excellence Centres (Federal, Provincial, Local)
AI Excellence Centres are meant to be Nepal's applied AI research and training backbone, established at the federal, provincial, and local levels, and housed inside universities, research institutions, and educational institutions rather than built as standalone government offices. The policy's stated goal is to have an AI Excellence Centre operating in every one of Nepal's seven provinces within five years.
Their responsibilities go well beyond research. According to the policy, AI Excellence Centres are tasked with running training and workshops to build AI skills among students and researchers, managing research data with an emphasis on privacy and ethics, advising the National AI Centre on policy and quality standards, supporting local researchers and institutions, and helping apply AI technologies across sectors like agriculture, health, tourism, and public service at the provincial and local level. They're also expected to help operate Nepal's planned AI Incubation Hub for startups and manage scholarships and internships for students building practical AI experience.
Who Oversees It: The AI Regulation Council
You may see this body referred to in different places as the "AI Regulation Council," the "AI Supervision Council," or even the "AI Regulatory Authority." These all refer to the same institution; the official name used in Nepal's National AI Policy 2025 document is the AI Regulation Council, and variations in naming largely come down to translation differences across media coverage.
The AI Regulation Council is chaired by the Minister for Communications and Information Technology, and its membership includes secretaries from the Office of the Prime Minister and Council of Ministers, the Ministry of Communications and Information Technology, the Ministry of Finance, and the Ministry of Education, Science and Technology, along with the chairperson of the Nepal Telecommunications Authority, the CEO of the e-Governance Board, three subject-matter experts from the private sector and universities, and the Chief of the National AI Centre, who serves as member secretary. The Council is required to meet at least twice a year, and its core job is providing guidance on fairness, transparency, and human rights protection in AI development, approving technical standards, and coordinating AI policy across all three levels of government. According to reporting from April 2026, the Council has since been formally established, though its practical impact will depend heavily on how actively it functions going forward.
Why This Matters for Startups and Researchers
For a Nepali AI startup today, the combination of a regulatory sandbox and provincial AI Excellence Centres could meaningfully lower the barrier to building and testing something new. Instead of needing to fully comply with every future AI regulation before you can even pilot a product, a sandbox structure would let a startup test with real users under temporary, supervised flexibility — similar to how Nepal Rastra Bank's fintech sandbox already lets young companies prove a concept works before chasing a full license.
For researchers, AI Excellence Centres represent a more concrete opportunity: access to institutional research infrastructure, training programs, scholarships and internships, and a direct channel to advise national policy through the National AI Centre. The policy also specifically mentions a "Brain Gain Program" intended to draw on the knowledge, skills, and capital of Nepali diaspora professionals working in AI abroad — a potential opening for NRN researchers and engineers who want to contribute to Nepal's AI ecosystem without necessarily relocating permanently.
Frequently Asked Questions
Is Nepal's AI regulatory sandbox operating yet?
The National AI Policy 2025 commits to creating a regulatory sandbox as a continuous action item, but detailed operational rules specific to AI had not been separately published as of this writing; Nepal Rastra Bank's already-finalized fintech sandbox offers the closest working model.
What is the difference between the AI Regulation Council and the National AI Centre?
The AI Regulation Council is the high-level oversight body that sets standards and guidance, while the National AI Centre is the operational secretariat that implements policy day-to-day and also serves as the Council's secretariat.
Where will AI Excellence Centres be located?
They are planned for federal, provincial, and local levels, based within universities, research institutions, and educational institutions, with a goal of covering all seven provinces within five years.
Discussion