Humanity Faces Extinction Threat From AI Bioweapons And Nuclear War As Scientists Identify 17 Catastrophic Global Dangers

Scientists Issue Chilling 2100 Extinction Warning As AI And Nuclear War Threaten Humanity's Survival

Humanity's Final Century? Scientists Warn Of Extinction By 2100

A Major Lancet Assessment Identifies Two Potential Extinction Threats And 15 Other Catastrophic Dangers, Warning That Decisions Made Today Could Determine Humanity's Future

Humanity could face extinction before the year 2100 through nuclear war or the deliberate misuse of artificial intelligence, according to a major international scientific assessment published in The Lancet.

The assessment, developed with expertise from 35 international commissioners, identifies 17 catastrophic threats to global health and human survival over the next 75 years. Two carry the potential for an even more devastating outcome: the disappearance of humanity itself.

The first is nuclear conflict capable of triggering widespread famine. The second is the malicious use of artificial intelligence, including its potential role in developing biological weapons.

Researchers stress that the likelihood of these extreme outcomes remains low. They have not predicted that humanity will become extinct, nor established that either catastrophe is inevitable.

Instead, their findings raise a different concern. Technologies and conflicts capable of causing unprecedented harm are developing alongside existing health, environmental and economic crises, while the world's ability to prevent them remains uneven.

The new findings were published on 10 October 2026 and presented at the World Health Summit in Berlin on 11 October.

Scientists Identify Two Threats That Could End Humanity

The Lancet Commission on 21st Century Global Threats to Health examined hundreds of possible dangers before selecting 17 with potentially catastrophic consequences between 2025 and 2100.

The work drew on historical evidence, forecasts and risk analysis covering 204 countries and territories.

Among the threats considered, nuclear conflict and malicious artificial intelligence stand apart because their possible consequences extend beyond widespread illness, economic collapse or mass casualties.

Under extreme circumstances, both could threaten the survival of the human species.

A nuclear war could disrupt the climate and global food production on an extraordinary scale. Deliberately weaponised artificial intelligence could potentially help hostile actors develop biological threats beyond the capacity of existing defences to contain.

The commission estimated that action to prevent nuclear conflict could avert approximately 133 million deaths and 9.35 billion disability-adjusted life years by 2100.

Measures reducing the risk of malicious AI could avert an estimated 10.4 billion disability-adjusted life years over the same period.

These figures are modelled estimates of potentially preventable harm across future scenarios. They are not predictions that those losses will happen, nor estimates of how many people will become extinct.

For rare events such as nuclear catastrophe and deliberate AI misuse, the researchers relied partly on structured expert judgement and specialist forecasting rather than historical data alone.

That makes the estimates especially uncertain.

Nevertheless, the assessment treats their possible consequences as sufficiently serious to justify preventive action.

How Nuclear War Could Cause A Global Famine

A nuclear war would produce destruction far beyond the cities and military facilities directly struck by weapons.

The immediate consequences could include blast injuries, intense heat, fires, radiation exposure, destroyed hospitals and the collapse of essential infrastructure.

Yet the longer-term consequences could become even more widespread.

Large nuclear exchanges could ignite extensive fires, sending smoke and soot high into the atmosphere.

If enough soot reached the upper atmosphere, it could reduce the sunlight reaching Earth's surface. Falling temperatures and changes to rainfall could then disrupt agricultural production across multiple continents.

Scientists refer to this possible climatic response as nuclear winter.

Crops could fail in countries never directly attacked. Food exports might collapse as governments attempted to preserve domestic supplies. International trade disruptions could intensify shortages in regions already struggling with hunger.

The result could be a global food emergency far larger than the immediate casualties from nuclear detonations.

The severity would depend on the number and type of weapons used, their targets, the amount of soot produced and the response of the climate system.

A limited nuclear exchange would not necessarily cause the same consequences as a large-scale conflict between major nuclear powers.

Nor does the existence of a possible nuclear winter establish that any nuclear war would eradicate every human being.

The commission's concern is that sufficiently extreme scenarios could produce consequences capable of threatening human survival on a planetary scale.

Artificial intelligence could add another danger to this situation.

As discussed in the risk of AI accidentally triggering World War Three, increasingly automated military systems could influence how quickly commanders interpret threats and respond during a crisis.

If those systems misinterpret information or encourage rapid escalation, human decision-makers could find themselves under even greater pressure.

The commission's nuclear assessment does not establish that AI will cause such a war. The connection illustrates how technological and military risks can overlap.

Artificial Intelligence Could Make Biological Weapons More Dangerous

The second extinction-level concern involves the malicious use of artificial intelligence.

This is different from the familiar science-fiction scenario in which an intelligent machine independently decides to destroy humanity.

The immediate concern is human misuse.

Artificial intelligence is becoming increasingly capable of assisting scientific research, analysing complicated biological information and accelerating work that would previously have demanded considerable specialist expertise.

These capabilities have legitimate medical applications.

Researchers can use AI to study proteins, identify potential treatments, analyse experimental results and support the development of new medicines.

But technologies that improve scientific productivity can also create opportunities for misuse.

The commission highlights the possibility that malicious actors could exploit AI to help develop dangerous biological weapons.

The concern is that future systems might lower the expertise, time or technical resources needed to undertake activities that currently present substantial barriers.

An AI-enabled biological threat would still have to overcome major practical obstacles. Designing a harmful biological agent on a computer is not equivalent to manufacturing, stabilising or spreading it successfully.

Laboratory facilities, physical materials, biological uncertainty and real-world constraints remain important.

There is no evidence that existing general-purpose AI systems can independently create a civilisation-ending biological weapon.

The warning concerns how those barriers might change as AI capabilities advance.

The distinction is explored in the wider debate over whether artificial intelligence could genuinely cause human extinction.

A catastrophe involving AI-assisted biological weapons would not require machines to become conscious or hostile towards humanity.

It could arise from people deliberately using advanced systems to pursue destructive objectives.

That places responsibility not only on software developers but also on governments, scientific institutions, laboratory security systems and international safeguards.

Seventeen Catastrophic Threats Facing The World

The commission's findings extend well beyond nuclear weapons and emerging technology.

Its wider assessment identifies 15 additional threats capable of causing enormous cumulative losses to human health.

These include climate change, conventional armed conflict, air pollution, antimicrobial resistance, naturally occurring pandemics, childhood malnutrition, obesity, smoking and harmful alcohol consumption.

Other threats include poor mental health, high blood pressure, unhealthy ageing, income inequality, low educational attainment and declining international development assistance for health.

The researchers assessed the impact of these dangers using disability-adjusted life years, commonly called DALYs.

A DALY represents one year of healthy life lost through premature death or time lived with illness or disability.

This measure allows researchers to compare very different problems using a common framework.

For the commission's assessment, a threat qualified as catastrophic if it had the potential to produce at least one billion lost healthy life years by 2100, could be assessed using available evidence and offered meaningful opportunities for intervention.

The framework deliberately captures threats that governments can do something about.

It also reveals a significant difference between catastrophic risks that already affect millions of people and more extreme dangers that remain uncertain.

Smoking, obesity and air pollution are established causes of health damage.

Human extinction through artificial intelligence or nuclear famine is a prospective outcome that has not occurred.

Placing them within the same framework does not imply that they have equal probabilities.

The purpose is to consider both the likelihood of harm and the scale of its consequences when deciding where prevention could make a difference.

The Health Consequences Could Reach Billions Of People

Some of the commission's largest projected health burdens come from familiar problems rather than speculative disasters.

Obesity alone was associated with approximately 23.1 billion potentially avoidable disability-adjusted life years by 2100.

The assessment estimated that improvements in education could address a potential burden of around 16.1 billion healthy life years, reflecting the powerful relationship between educational attainment, family circumstances and long-term health.

Income inequality was associated with an estimated 13.8 billion potentially avoidable healthy life years.

These figures describe modelled opportunities to reduce future harm, not forecasts that the entire burden will necessarily occur.

They also challenge the idea that catastrophic global threats must always involve a sudden event.

A health crisis unfolding gradually across generations can impose enormous cumulative damage without a single moment of collapse.

The geographical effects would be uneven.

Wealthier regions face substantial burdens from non-communicable diseases, including obesity and mental health conditions.

In lower-income regions, limited access to education, childhood malnutrition and fragile health systems can contribute to different patterns of lost healthy life.

The commission also examined the consequences of reductions in international health assistance.

Sub-Saharan Africa is expected to face a particularly large share of the preventable health burden associated with a major collapse in development assistance.

These disparities matter because the countries least equipped to respond to severe disruption may also be among those most exposed to its consequences.

Climate Change Could Intensify Existing Threats

Climate change forms another major part of the assessment.

The commission compared future scenarios in which global warming is limited to around 1.8 degrees Celsius with scenarios reaching approximately 4.4 degrees Celsius.

Its modelling suggested that avoiding the higher-warming pathway could prevent around 1.89 billion healthy life years being lost through measurable consequences, including extreme heat and air pollution.

The researchers cautioned that this estimate does not capture every possible impact.

Displacement, drought, mental health consequences and wider social disruption can be difficult to quantify accurately over such long periods.

Climate change also interacts with other risks.

Extreme heat can place pressure on healthcare systems. Agricultural disruption can worsen food insecurity. Environmental changes may increase the difficulty of preventing disease in vulnerable communities.

Political instability and economic hardship can further weaken a country's ability to respond.

This is why the commission argues against treating each global threat as an isolated problem.

A severe crisis can reduce the resources available to prevent the next one.

The danger is not simply the scale of individual threats, but the possibility that several could emerge or intensify at the same time.

How Reliable Are Predictions About Human Extinction?

Predicting events 75 years into the future involves enormous uncertainty.

Researchers can estimate future population trends, disease burdens and environmental changes using historical data and mathematical models.

But rare catastrophic events present a different challenge.

There is no historical record of humanity becoming extinct through nuclear war or AI-developed biological weapons.

Scientists cannot calculate their probabilities with the same confidence as they can estimate the health effects of a well-studied disease.

Instead, researchers must consider possible mechanisms, available evidence, expert assessments and assumptions about how technology and human behaviour might develop.

The commission recognises these limitations.

Some of its risk assessments use structured expert judgement and superforecaster estimates, particularly where sufficient historical observations do not exist.

These methods can help decision-makers examine uncertain futures, but their results remain dependent on the assumptions and information available at the time.

Importantly, the commission does not present a fixed ranking of threats or a single inevitable future.

Its assessment is intended to change as new evidence becomes available.

That is also important when considering the growing debate over the potential for AI systems to become more capable than their human operators.

Warnings about existential risk do not establish that an extinction event is approaching.

They identify possible failure pathways whose consequences may justify action even when the probability remains uncertain.

Scientists Say Governments Still Have Time To Act

The commission does not argue that the future is beyond human control.

Its central recommendation is that governments should devote more resources to anticipating, preventing and preparing for major threats before they become unmanageable.

The proposed approach has three principal elements.

First, countries should build health systems capable of responding to multiple types of crisis. That means developing lasting capacity for forecasting, preparedness, prevention and emergency response.

Second, governments and institutions should accelerate investment in useful innovations, including vaccines, antibiotics, treatments for widespread diseases and technologies that reduce environmental harm.

Third, emerging risks should be monitored regularly and independently so that decisions reflect changing evidence.

The commission proposes continued assessment through the Lancet Monitor of 21st Century Global Threats to Health.

Its co-chair, Professor Christopher Murray of the Institute for Health Metrics and Evaluation at the University of Washington, emphasised that leaders still have the power to influence the outcome.

Murray said the choices governments make today will shape the health and wellbeing of the next generation.

He also warned that delaying preventive action would increase the costs passed into the future.

The recommendations extend beyond health ministries.

Nuclear safety involves defence and diplomacy. Artificial intelligence safety involves technical oversight, scientific research and security controls. Climate policy depends on energy systems, economics and international cooperation.

Many of the proposed solutions therefore require decisions across institutions that do not traditionally work together.

The World Faces A New Test Of Global Preparedness

The new assessment arrives at a time when governments are being asked to manage established health problems alongside technologies whose future capabilities remain difficult to predict.

Artificial intelligence offers opportunities to improve medicine, research and productivity, yet also creates new questions about access, accountability and potential misuse.

Nuclear weapons have existed for decades, but their destructive potential remains extraordinary.

Meanwhile, climate change, disease, inequality and fragile international health funding create pressures that can weaken societies long before an exceptional catastrophe occurs.

The commission's findings suggest that preventing future disasters requires more than responding effectively after an emergency begins.

It requires institutions able to recognise dangerous developments early and change course before damage becomes irreversible.

The research team plans to update its assessment annually, incorporating new evidence and emerging threats.

Those updates will provide an opportunity to examine whether today's risks are changing and whether governments are making progress in reducing them.

The year 2100 is not an established deadline for human extinction.

It is the horizon against which the commission has assessed the consequences of decisions being made now.

Sources

• The Lancet — The Lancet Commission On 21st Century Global Threats To Health — Original scientific commission, published 10 October 2026, examining catastrophic risks to global health and human survival.

• World Health Summit — New Lancet Commission Provides Roadmap To Protect Global Health And Human Survival — Official 11 October 2026 announcement detailing the commission's methodology, projected health impacts, recommendations and uncertainty.

• The Guardian — Nuclear War And Malicious Use Of AI Threaten End Of Humanity By 2100, Says Lancet Assessment — Reports the extinction-risk findings, nuclear famine scenarios and comments from commission members.

Next Reads

• AI Doomsday Explained: How Real Is The Risk Of Human Extinction? — Examines the scientific arguments behind AI existential risk, machine autonomy and the possibility of humans losing control.

• AI Could Trigger World War Three — And Humanity May Have Minutes To Stop It — Explains how military AI, false warnings and accelerating decision-making could increase the danger of nuclear escalation.

• Anthropic's Terrifying AI Warning: The Technology It Is Building Could Threaten Humanity Itself — Investigates why a leading AI developer has publicly acknowledged the possibility of catastrophic or existential consequences.