Nuclear war and the misuse of AI could threaten the very survival of humanity. Yet everyday problems such as obesity and educational inequality can also cause enormous health losses over the long run. Comparing threats of such different natures on a common scale has never been easy.
On October 11, 2026, a panel of experts convened by the medical journal The Lancet presented its report, "The Lancet Commission on 21st Century Global Threats to Health", at the World Health Summit 2026 in Berlin, Germany. Based on analysis by the Institute for Health Metrics and Evaluation (IHME) at the University of Washington, the report assesses health risks that could arise over the 75 years from 2025 to 2100 across 204 countries and territories.
The commission classified 17 items as "catastrophic threats," ranging from existential risks such as nuclear conflict and the malicious use of AI to chronic problems such as obesity, low educational attainment, and income inequality. It then used mathematical models to estimate how much health loss could be avoided through countermeasures.
The report, however, is not a definitive forecast of future deaths or disease incidence. The figures compare the best and worst plausible scenarios, estimating how much health loss could be avoided with appropriate action.
Understanding that distinction is key to reading the report correctly.
データを表で見る
| Avoidable cumulative DALYs (billion DALYs) | |
|---|---|
| Obesity measures | 23.1 |
| Better access to education | 16.1 |
| Reducing income inequality | 13.8 |
| Countering malicious AI use | 10.4 |
| Preventing nuclear conflict | 9.35 |
| Limiting climate change | 1.89 |
The estimates suggest that problems deeply rooted in daily life, such as obesity and social inequality, could cause health losses equal to or greater than those from nuclear conflict or catastrophic AI-related events.
How were the 17 catastrophic threats chosen?
The commission set quantitative criteria for what counts as a "catastrophic" threat.
The commission was co-chaired by Professor Christopher J. L. Murray, director of IHME, and Dr. Natalia Kanem, former Executive Director of the United Nations Population Fund (UNFPA). Thirty-five experts in health, economics, security, governance, technology, and other fields took part and considered hundreds of potential threats.
Three conditions were used to select the final 17.
First, the threat could cause at least 1 billion DALYs (disability-adjusted life years) of health loss by 2100. Second, sufficient data existed to evaluate it with mathematical models. Third, there was room to reduce the harm through policy changes, international cooperation, technological innovation, and similar measures.
These criteria are also explained in the official announcement published by the World Health Summit.
The DALY is a public health metric that adds together years of healthy life lost to illness or disability and years of life lost to premature death.
One DALY equals one year of healthy life lost. It captures effects that simple death counts miss, such as declining health from chronic disease and the impact of early deaths among young people.
The 17 selected threats fall into two broad groups.
The threats identified as potentially the most severe are nuclear conflict and the malicious use of AI. The commission classified both as low-probability events that could, in the worst case, lead to human extinction.
The remaining 15 items include air pollution, antimicrobial resistance (AMR), childhood undernutrition, climate change, armed conflict with conventional weapons, declining development assistance for health, excessive alcohol consumption, high blood pressure, income inequality, low educational attainment, naturally occurring pandemics, obesity, mental health problems, smoking, and aging with health problems.
These are not unrelated issues. Income inequality and lack of educational opportunity, for example, affect health, while climate change and conflict can worsen existing vulnerabilities.
Nor is the list of 17 a fixed ranking of importance. A related editorial in The Lancet assumes that assessments will change as new research findings and observational data accumulate.
The commission's aim is not to rank crises once and be done, but to build a mechanism for continually reviewing priorities for action based on scientific evidence.
How much harm could nuclear conflict and AI misuse cause?
Among the 17 threats, nuclear conflict and the malicious use of AI hold a special position.
Both are considered unlikely to occur, but the damage would be extremely large if they did.
The commission estimates that measures to prevent nuclear conflict could avoid about 133 million deaths and 9.35 billion DALYs of health loss by 2100.
This is not a prediction that nuclear war would necessarily kill 133 million people. It is an estimate of the health harm that could be avoided through countermeasures, taking into account both the likelihood of nuclear conflict and the scale of damage expected if it occurred.
Meanwhile, countermeasures against the malicious use of AI are estimated to avoid 10.4 billion DALYs of health loss.
There are several scenarios in which AI could threaten human health and safety.
One is its use in designing biological weapons or in technologies that make pathogens more dangerous. There is also concern that AI could be used to attack critical infrastructure and healthcare systems.
Serious malfunctions of AI systems and loss of human control over increasingly advanced AI are also considered AI-related risks in a broad sense. However, the malicious use of AI by people and the malfunction or loss of control of AI itself differ in their causes and in the countermeasures they require.
A key point in interpreting these figures is the difference in estimation methods.
For nuclear war and catastrophic events caused by advanced AI, there are no past observational data sufficient to predict future probabilities and damage with precision.
For common infectious diseases or air pollution, future trends can be estimated from past disease statistics and environmental data. For events that threaten human survival, the same approach cannot simply be applied.
The commission therefore used other methods, including Structured Expert Judgement and assessments by forecasting experts.
In these approaches, experts' judgments about probabilities and damage scales are organized in a defined way and incorporated into mathematical models.
The results, however, depend heavily on expert judgment and underlying assumptions. They are not as certain as estimates from conventional epidemiological models.
In its methodological explanation in the official announcement, the commission states that threats with scarce historical data, such as nuclear conflict, future pandemics, and AI misuse, carry large uncertainty and that the figures should be read as a guide.
The figures of 9.35 billion DALYs for nuclear conflict and 10.4 billion DALYs for AI misuse therefore do not indicate that these crises will certainly occur. They are estimates for considering the scale of potential harm and the value of prevention.
Obesity causes the greatest health loss; educational and income inequality also serious
While nuclear conflict and AI catastrophes attract attention, chronic problems such as obesity and social inequality occupied the top places in the estimates of avoidable health loss.
The largest was obesity, with 23.1 billion DALYs of health loss potentially avoidable by 2100.
Obesity raises the risk of many diseases, including cardiovascular disease and diabetes. As dietary changes and physical inactivity spread worldwide, long-term health harm could differ greatly depending on how widely prevention and treatment are adopted.
The figure of 23.1 billion DALYs is the largest among the 17 threats the commission evaluated.
Improving access to education could also avoid a large health loss, estimated at 16.1 billion DALYs.
The greatest impact was expected among women and children in sub-Saharan Africa.
As educational attainment rises, people gain more opportunities to learn about health and hygiene, and nutrition and access to healthcare tend to improve. Higher educational attainment among mothers is also linked to lower child mortality.
The commission points out that the spread of education can bring not only direct health improvements but also, over the long term, lower population growth rates.
Reducing income inequality could avoid an estimated 13.8 billion DALYs of health loss.
In its official announcement, the commission explains that this scale is equivalent to about 25 COVID-19 pandemics.
This does not mean deaths would be 25 times as high. It is a comparison that converts years of healthy life lost to illness, disability, and premature death into DALYs.
For climate change, reducing greenhouse gas emissions is estimated to avoid 1.89 billion DALYs of health loss.
This figure compares a scenario in which global average temperature rises by about 4.4°C above preindustrial levels with one in which the rise is held to about 1.8°C.
It covers direct effects that can currently be quantified, such as health harm from extreme heat and some air pollution associated with rising temperatures.
The commission cautions, however, that the 1.89 billion DALY estimate does not represent the whole of climate change's harm.
It excludes indirect harms that are hard to quantify, such as food shortages caused by drought, displacement from sea-level rise and disasters, and mental health effects of lost livelihoods.
Actual health harm from climate change could therefore exceed this estimate. If greenhouse gas emissions are not cut, the effects will continue beyond 2100 and could worsen other health risks.
Another serious problem identified is the decline in development assistance for health (DAH) from wealthy countries to low- and middle-income countries.
If aid were sharply reduced or halted, up to 1.5 billion DALYs of additional health loss could result by 2100, with about 70% expected to be concentrated in sub-Saharan Africa.
The less developed a region's healthcare and social security systems, the greater the impact of reduced aid. Future disease burden will also vary with population growth and economic development.
Problems such as lack of education, poverty, and income inequality do not affect health only on their own. They also amplify the harm caused by infectious diseases and climate change.
| Threat | Health loss avoidable by 2100 | Main estimation method | Interpretive caveats |
|---|---|---|---|
| Obesity | 23.1 billion DALYs | Comparison of epidemiological data and intervention scenarios | Depends on how widely prevention and treatment spread |
| Low educational attainment | 16.1 billion DALYs | Demographic models and analysis of links to health | Depends on where and how fast educational access improves |
| Income inequality | 13.8 billion DALYs | Analysis of links between income inequality and disease burden | Uncertainty about causality and the effects of policy intervention |
| Malicious use of AI | 10.4 billion DALYs | Structured expert judgement, among others | Historical observational data are scarce; uncertainty is extremely large |
| Nuclear conflict | 9.35 billion DALYs; about 133 million deaths | Expert judgment and damage models, among others | Depends heavily on assumed probability and scale of damage |
| Climate change | 1.89 billion DALYs | Comparison of temperature-rise scenarios | Many indirect harms, such as displacement and food shortages, are not included |
| Sharp reduction or halt of development assistance for health | Up to 1.5 billion DALYs | Model combining changes in aid with health effects | About 70% of harm expected to be concentrated in sub-Saharan Africa |
Figures are based on the Lancet Commission's report and official announcement. All are scenario-based estimates for 2025–2100, not confirmed future harm. Health harms across items may overlap, so they cannot simply be summed.
The estimates are not prophecy but the size of harm policy can change
The most important point in reading this report is not to confuse its figures with definitive predictions of the future.
What the commission estimated is the difference in health loss between best-case and worst-case scenarios. This metric, called "Amenable DALYs," represents health harm that could potentially be avoided through policy intervention and other measures.
IHME's analysis also includes a reference scenario in which current trends continue.
The worst-case scenario does not necessarily mean only a case in which authorities take no action at all. It is set to examine how much health harm could vary depending on differences in future social conditions, technology, and policy.
Not all of the estimated losses will therefore necessarily occur, and there is no guarantee that the best-case scenario can be achieved.
The 17 threats are also interrelated.
For example, loss of farmland and water shortages caused by climate change could raise the risk of regional conflict. If conflict destroys medical facilities and sanitation, it becomes harder to contain outbreaks of infectious disease.
One crisis can worsen another, and harm could spread in a chain reaction.
Professor Murray also stresses that the 17 threats are interconnected and cannot be dealt with in isolation.
However, the individual estimates presented do not incorporate all of these complex chains into the mathematical models.
In the analysis, health harm is in part estimated threat by threat, and the effect of multiple threats overlapping is not necessarily fully captured. Because health harms may also overlap, the DALYs for the 17 items cannot simply be added up and treated as the world's total loss.
Attention must also be paid to where harm occurs and which people are affected.
Global aggregate figures do not apply uniformly to residents of every country, including Japan.
For example, health harm from the decline in development assistance for health is expected to be concentrated in sub-Saharan Africa.
Even within a single country, people who lack adequate access to healthcare and education, or who are in economic hardship, may be hit harder by crises.
Nor does the report prove through clinical trials the effectiveness of any particular drug or technology.
Rather, it offers a framework for thinking about which challenges limited public funding and research and development investment should be directed toward in order to prevent the most future health harm.
Three action plans proposed by the commission
To address the 17 threats, the Lancet Commission sets out three courses of action to pursue over this century.
1. Build readiness
The first pillar is to put in place systems for forecasting, prevention, and preparedness in ordinary times, rather than responding only after a crisis occurs.
The commission calls on countries to build into their healthcare systems expert teams that continually monitor a range of health threats and handle prevention and emergency response.
On funding, it proposes shifting from securing ad hoc budgets each time a disaster or outbreak occurs to preparing long-term, stable sources of financing.
It is particularly important that measures reach resource-poor regions and people who are easily left behind.
If preparedness is concentrated in wealthy countries and regions, global health harm cannot be adequately reduced.
2. Back breakthroughs
The second pillar is to expand investment in research and development for threats that existing measures cannot adequately address.
Specific examples include vaccines enabling rapid response to emerging infectious diseases, new antimicrobials effective against drug-resistant bacteria, safer and more accessible obesity treatments, and climate-related technologies that help cut greenhouse gas emissions.
Developing new technologies alone is not enough, however.
If drugs and preventive technologies reach only some wealthy countries, their effect in reducing global health loss will be limited.
Alongside investment in innovation, the commission says international systems and cooperation are needed so that the benefits reach low- and middle-income countries equitably.
3. Track threats
The third pillar is to build a mechanism for regularly reassessing future health risks and using the results in policy decisions.
The size and likelihood of threats will change with the emergence of new diseases and technologies and with shifts in social conditions.
The 17 items should therefore not be treated as fixed; assessments need to be continually updated to reflect the latest data and research.
The commission proposes creating an independent monitoring body, the "Lancet Monitor of 21st Century Global Threats to Health," to fill this role.
The aim is to deliver the latest scientific findings not only to health authorities but also to policymakers responsible for finance, security, climate policy, and other areas.
Common to all three proposals is the view that health threats cannot be solved by the medical sector alone and must be tackled together with economic, security, and environmental policy.
Humanity's future can be changed by policy choices
Dr. Richard Horton, editor-in-chief of The Lancet, has praised the report as offering a framework for responding to more complex and wide-ranging threats, moving beyond the conventional approach of treating the crises humanity faces as separate problems.
Co-chair Dr. Kanem also noted that recent global events have shown how fragile social stability is and how heavy a burden it is to respond only after a crisis occurs.
She also stressed that the future is not predetermined and can be changed by the policy decisions made today.
What the report makes clear is that catastrophic crises such as nuclear war and the misuse of AI are not the only things that will shape humanity's future.
Problems that already exist in society, such as obesity, educational inequality, and income inequality, could also lead to extremely large health harm over the long term.
Of course, not all threats can be predicted with the same precision. For catastrophic events with little history of occurrence in particular, the estimates carry large uncertainty.
Still, there is value in comparing crises from different fields on a common scale and showing where room for action exists.
The report does not predict how much health humanity will lose by 2100. It asks what policies we should choose now, and where we should invest resources, to reduce future health harm.
From preventing nuclear conflict to tackling obesity and expanding educational opportunity, what these different challenges share is the importance of acting before a crisis becomes severe.