North Korea AI Job Risk 2026: 7.6M Workers, Craft Trades at 42.2%

7.6M workers in a closed planned economy

Data transparency statement: Every number in this analysis is an ILO statistical model estimate. The ILO produces employment estimates for all UN member states including North Korea, but these are derived from demographic data (population, age structure, historical patterns) combined with economic modelling - not from national labour force surveys or administrative records that the DPRK provides. North Korea's actual employment structure is one of the least known in the world.

What is known from defector accounts, satellite imagery analysis, and South Korean government assessments is that North Korea operates a state-directed planned economy where workers are assigned to workplaces by the state. The distinction between employed and unemployed does not function as it does in market economies - workers are assigned, not hired. The economy is organized around state enterprises, cooperative farms, military units, and the Korean Workers' Party's extensive organizational apparatus.

North Korea's economy went through a partial marketization in the 1990s-2000s following the collapse of the Soviet Union (which ended DPRK subsidies), the devastating 1994-1998 famine (Arduous March), and the partial emergence of informal markets (jangmadang). By the 2020s, informal markets were a significant source of household income alongside state work assignments. The ILO model cannot easily capture this hybrid structure.

Why craft trades at 42.2% - and why AI is irrelevant

The ILO model estimates 42.2% of North Korean workers in the craft and related trades category - the highest craft share in this analysis batch. This is structurally consistent with what outside analysts believe about North Korea's economy: a Soviet-style industrial structure that prioritized heavy industry, mining, manufacturing, and military production. Wonsan, Hamhung, Hungnam, and Chongjin are industrial cities with steel plants, chemical factories, and shipyards that would employ large numbers of craft and trades workers.

The 8.5/10 AI score for clerical workers and 6.5/10 for professionals is entirely theoretical for North Korea. The reason is the Kwangmyong intranet system: North Korea has a nationwide internal intranet that ordinary citizens can access at designated terminals, but it is entirely closed to the global internet. No commercial AI tools - ChatGPT, Copilot, Gemini, or any equivalent - are accessible through Kwangmyong. The state controls all information flow. The concept of a worker independently adopting an AI productivity tool is structurally impossible in this system.

The DPRK's own research institutions work on computing and technology, and North Korea has produced sophisticated cyberweapons (attributed to Lazarus Group and other state-sponsored hacking units). But this is state-directed offensive cyber capability, not civilian AI adoption. The workers in the modelled dataset have no access to and no exposure to AI tools in any meaningful sense.

* AI scores represent theoretical occupation-level exposure. Practical AI exposure for all North Korean workers is zero due to the absence of civilian internet and AI infrastructure.

The safest jobs in North Korea

In theoretical model terms, craft and trades workers at 2.5/10 covering 3.21M workers (42.24%) are the safest from AI. Agricultural workers at 2.0/10 cover 1.16M workers (15.23%), reflecting the cooperative farming system. Elementary occupations at 2.0/10 cover 786,440 workers (10.35%). In practice, all North Korean jobs are safe from AI in the current system - but for the wrong reason (structural impossibility of AI access) rather than the right reason (skill resilience against automation).

The state-directed character of North Korean employment means that even if AI tools were theoretically available, the decision to deploy them would rest with state enterprise directors and party cadres, not with individual workers or competitive market pressure. The mechanism that drives AI adoption in market economies - firms competing to reduce costs and improve output - does not operate in the DPRK's system in the same way.

What this means for North Korea workers

For North Korean workers, the more immediately relevant technology question is not AI but access to information. The state's control over information - enforced through Kwangmyong, state media, and the strict criminalisation of unauthorized media consumption - means that North Korean workers have no awareness of AI developments in the outside world. The economic questions that matter to them are food security, the allocation of work through state enterprise assignments, and the supplementary income from jangmadang market activity.

There is, however, a scenario worth noting: North Korean IT workers operating abroad. The DPRK sends technology workers to China and other countries under strict supervision, where they develop software for foreign clients to generate hard currency for the regime. These workers have internet access in their host countries and have demonstrably used advanced computing tools. This offshore IT workforce - estimated by South Korean and US government sources to number in the thousands - is the only North Korean population with actual AI tool access. But they are not in the ILO domestic employment model.

Any scenario in which AI becomes relevant to North Korean domestic workers would require fundamental political change - either a policy shift under the current regime (possible but historically unlikely) or regime transformation. Given that context, this analysis is better understood as a historical and structural document than as a forward-looking risk assessment.