Last month, Nepal took government systems offline to patch two critical PHP flaws, a reminder that sovereignty starts with knowing what we run and being able to change it. The stakes extend well beyond government servers. Technological advancements have reshaped how we live, interact and communicate. The most pervasive technology of our times is AI, which is now integral to business, administration, innovation, and almost all human function. Its importance can be gauged from the fact that today the question for countries worldwide is not whether to implement AI but to implement in a way to serve national interests without compromising autonomy and sovereignty. 

AI in Nepal can be a gateway to progress and prosperity, ensuring equitable access to education and enabling businesses, specifically small and mid-sized enterprises, to compete in the global market. This said, technology needs to be implemented responsibly in a way that the country retains its sovereignty and doesn’t become subservient to foreign-controlled data and systems. Technological independence is therefore extremely important. This does not imply building technology from scratch but forging foreign partnerships that help build local capabilities without encroaching on sensitive data and systems. Thus, Nepal must be able to govern the AI it buys—it must retain the ability to choose, govern, and adapt technologies to align with its unique needs while also having complete control over data and decision-making.

Building capacity, securing control

The country has already taken concrete steps in this direction, the foremost being introducing the National Artificial Intelligence Policy, 2025, which sets the guidelines for AI use in Nepal, with the aim of leveraging technology for research, innovation, and business and administrative support, while addressing all data and related issues to promote national interests. It has also received a $50 million loan from the World Bank to support the Nepal Digital Transformation Project, and another $90 million from the Asian Development Bank for strengthening the digital infrastructure needed to effectively implement the technology.

However, these investments must not undermine Nepal’s ability to independently control, monitor and audit its AI systems. This is in line with the OECD's Digital Government Outlook 2026, which reiterates that public institutions must ensure accountability and transparency, maintain data rights and manage lifecycle risks. Thus, to manage the technology it buys, Nepal must develop local talent, institutional capacities, and frameworks to retain control and safeguard its technological autonomy.

The need for technological autonomy

Technological autonomy is crucial in that while AI may improve efficiency and save time, integrating it with proprietary software may expose sensitive data to external entities, posing a significant data breach. Thus, while acquiring AI technology, it’s extremely crucial to address relevant questions, more specifically in terms of who retains control over records, who audits the systems, and what contingency measures are in place in case the service is compromised or becomes unavailable. These concerns must be addressed before and not after technology acquisition and implementation because it’s less expensive to fix issues at the outset than when they are entrenched in the systems.

Preserving Nepal’s rich heritage

Data security and technological autonomy apart, the new AI systems must align with Nepal’s rich cultural and linguistic heritage. While most global models adapt well to Nepali, performance may vary across dialects and even some public sector contexts. This problem can be resolved by investing in translation resources, Nepali-language datasets, digitised public archives, and evaluation benchmarks. Domestic companies and Universities too can build the necessary tools that preserve our rich heritage and independently evaluate if imported systems meet local needs without simply relying on the vendors’ claims. For instance, Kathmandu University’s language lab has undertaken a project to build a Nepali- Tamang and English translation system and benchmarked global multilingual models on its own data instead of relying on claims.

Building capabilities

First, Nepal needs to build its digital infrastructure, data, and skills, aligned with the UNCTAD's Technology and Innovation Report 2025, which sets the guidelines for public institutions on how to govern and share data through external systems, and who is accountable for its use. Second, the country must build domestic skills and capabilities to reduce dependence on external vendors to preserve national autonomy while also ensuring preservation of its rich cultural and linguistic heritage. Third, contracts with overseas vendors should include provisions for interoperability, auditability, data portability, exit options and future maintenance expenses. Although relying on one vendor may lower upfront costs, it could become expensive over time by reducing both switching options and bargaining power. Fourth, there is a need for clear safeguards for agentic AI with restricted access, audit logs, and designated approval for major decisions. Before deployments, pilot systems should first be tested for accuracy, security, bias, and failure risks. Fifth, there is also a need to appropriately measure outcomes in terms of the use of domestic talent and open Nepali-language resources to gauge whether the country is actually retaining technological autonomy or simply producing documents.

Sovereignty is not isolation

It is important to note that technological autonomy does not have to be at the cost of isolation, especially in an interconnected world where even advanced economies are interdependent on each other. The way forward is to collaborate with external stakeholders, including Universities and technological firms in ways to build and safeguard and not replace domestic talent and expertise. The choice is not whether to adopt or resist AI—the winds of change are already reshaping economies and societies worldwide, and Nepal is no exception. The more important aspect is to build local capabilities, expertise, and institutions needed to engage with the global AI ecosystem to maintain our autonomy. Rather than remaining dependent on foreign technologies, Nepal must emerge as an equal global partner supporting innovation, while safeguarding its own national interests and technological sovereignty. Nepal has the policy, financing and institutional support to make this happen; it should not let it pass.