The Lancet Commission on 21st Century Global Threats to Health, co‑chaired by Christopher J. L. Murray (director of the Institute for Health Metrics and Evaluation, IHME), modelled 17 major global health threats across 204 countries and territories for 2025–2100. Researchers analysed hundreds of diseases and socio‑economic drivers, then placed 17 priority risks on the top two levels of a four‑tier “threat pyramid.”
The report was not intended as an apocalypse prediction but rather as a common metric for policymakers: how many healthy life years could be saved with better policies over the next 75 years.
Method — what is a DALY and what does the report measure?
The commission used the expected cumulative amenable disability‑adjusted life‑year (DALY) metric. One DALY represents one year of life lost to premature death or lived with disability. The large billion‑scale figures in the report do not show total guaranteed destruction; they represent the difference between a plausible worst and best scenario derived from historical country data — that is, the policy‑avoidable margin.
The metric is an expected value: it multiplies the probability of an event by its estimated damage. That explains why events of low probability but catastrophically large impact (such as nuclear war or a global AI disaster) can produce very large expected DALY burdens.
Why are nuclear war and AI at the pyramid peak?
The commission placed two threats at the pyramid summit: nuclear conflict and artificial intelligence. Both are classified as having extinction potential because, even with low probabilities, they could trigger civilizational collapse and therefore produce enormous expected losses.
- For nuclear conflict the report estimates 9.35 billion avoidable DALYs and 133 million avoidable deaths. The modelling does not assert that exactly that number of deaths would occur; rather, it reflects the expected value produced by multiplying a very large potential impact by a low probability of occurrence.
- For AI‑related risks the commission assigns 10.4 billion DALYs. The analysis considers three distinct channels: (1) AI‑facilitated design of biological weapons; (2) paralysis of critical infrastructure via AI‑based cyberattacks, AI hallucinations in clinical guidelines, mass political deepfakes, or attacks on autonomous/nuclear systems; and (3) loss of control — emergence of an unaligned agentic AGI (general artificial intelligence) that acts autonomously with goals not aligned to human interests.
Nuclear risk also draws on historical near‑misses: between 1948 and 2010 there were 22 nuclear weapon accidents and 21 close‑call incidents, and current great‑power tensions and changes in some nuclear doctrines increase contemporary risk. A major exchange would, beyond immediate blast casualties, loft soot into the upper atmosphere, causing years of cooling and agricultural collapse. The commission highlights that the largest portion of the expected health burden from nuclear famine (2.97 billion DALYs and 42.4 million deaths) would fall on sub‑Saharan Africa — a region without nuclear weapons but highly vulnerable due to youthful populations and food insecurity.
Because there are no historical time series for global AI catastrophes, the commission relied on forecasting exercises (including results from the Forecasting Research Institute), domain experts and so‑called superforecasters. Superforecasters’ median estimates put the chance of an AI catastrophe (defined as deaths of at least 10% of global population) by 2100 at 2.1% and the chance of total human extinction at 0.4%; surveyed AI researchers provided higher medians (12.0% and 3.0% respectively).
The commission stresses that while infrastructure and bio‑risks can be mitigated by regulation (e.g., mandatory screening of DNA synthesis orders, stronger cyber‑security), loss of control over an AGI is poorly addressable by conventional policy tools.
The largest expected health losses — slow, large burdens
Although wars and technological risks dominate headlines, over the 75‑year horizon the largest avoidable health losses stem from slow social, economic and metabolic processes:
- Obesity, poor diet and physical inactivity: 23.1 billion avoidable DALYs (the top risk in 154 of 204 countries). The best scenario assumes a 50% global uptake of new generation (GLP‑1) weight‑loss drugs.
- Low education: 16.1 billion DALYs. Educating women reduces both child and adult mortality and lowers fertility; part of the DALY gain derives from a smaller future population.
- Income inequality: 13.8 billion DALYs.
- Poor mental health: 9.35 billion DALYs (already the largest burden among 10–24 year olds).
- Unhealthy ageing: 8.26 billion DALYs — healthy years are increasing more slowly than total life expectancy, so people will on average spend nearly three more years living with chronic conditions late in life.
These “silent killers” will place major strains on pension and health financing systems in the latter half of the century.
Climate, pandemics and cuts to international aid
The commission attributes 1.89 billion avoidable DALYs to the difference between a high‑emissions 4.4 °C pathway and a lower 1.8 °C pathway. The calculation primarily captures direct heat stress and chronic disease from PM2.5 pollution due to fossil fuels and wildfires; it does not comprehensively include hard‑to‑model consequences such as mass climate migration, sea‑level rise, droughts or interstate conflict, which could amplify systemic risk.
Natural‑origin pandemics were modelled using a historical database of 61 major epidemics reaching back to 1600. While frequency of moderate epidemics has declined since the 1950s, the occurrence rate of catastrophic pandemics (scaled to today’s population, causing millions of deaths) has not fallen.
The report also quantifies the health cost of geopolitical and budgetary shifts: cuts in development assistance for health (DAH) could produce 1.50 billion avoidable DALYs. It notes that in 2025 several major donor countries reduced ODA to increase defence spending; such cuts hit vaccination, malaria and HIV programmes and translate rapidly into lives lost in low‑income, aid‑dependent countries.
Regional differences — what threatens Central and Eastern Europe?
Threat priorities differ substantially by region. In sub‑Saharan Africa low education, income inequality and conventional wars dominate. In the Central Europe, Eastern Europe and Central Asia super‑region obesity is the top risk (967 million avoidable DALYs, 9.1% of the region’s reference disease burden), followed by AI risks (404 million DALYs), poor mental health (361 million), nuclear conflict (351 million), smoking (348 million) and unhealthy ageing (334 million).
Recommendations — what should be done now?
Murray and colleagues conclude that future health losses are not inevitable but postponing action raises both societal and fiscal costs. The commission’s headline recommendations include:
- Rebuilding social trust and functional international governance.
- Building resilient food systems.
- Three policy packages: threat‑prepared health systems (with dedicated foresight units, strong primary care and prevention budgets embedded in national budgets); radically increased funding for breakthrough innovation (from pan‑virus vaccines and new antibiotics to cheaper oral obesity drugs, fusion energy and atmospheric CO2 removal); and strict international regulation of cutting‑edge AI models and biological design tools.
The commission proposes an annually updated Monitor of 21st Century Global Threats to Health whose findings would be directed not to specialist health conferences but to finance ministers and national security decision‑makers (for example through G20 finance and health working groups, the Munich Security Conference, the World Economic Forum and the UN General Assembly).
Conclusion
The Lancet commission provides a clear compass for economic and security policy: highlighting AI and nuclear conflict is warranted because, even at low probability, they threaten civilizational outcomes. Yet the 21st century’s health and economic balance will be decided by whether societies can simultaneously contain technological extremes and commit substantial resources to preventing demographic, environmental and chronic disease risks. The avoidable DALY burdens presented by the modelling underline that timely policy choices can save very large numbers of healthy life years.
