17 threats, 75 years, one number labelled extinction
A Lancet commission costs 17 global health threats in lost years, and two could end humanity by 2100.
By The Ledger · · 5 min read
Two of the 17 threats on a new global health list carry a risk the other 15 don't: human extinction by 2100. The Lancet Commission on 21st Century Global Threats to Health, publishing its first assessment, names malicious use of artificial intelligence and nuclear conflict as the only risks capable of ending the species outright, not merely shortening it. The other 15 threats, from air pollution to aid cuts, are measured in years lost. These two are measured in whether anyone is left to count them.
A sub-Saharan health worker already knows what 1.5 billion lost years feels like in a clinic
The commission's method converts abstract risk into disability-adjusted life years, or DALYs — a measure of healthy years lost to sickness, disability or early death. By that measure, the collapse of international health aid could cost up to 1.5 billion DALYs by 2100. Nearly 70% of that falls on sub-Saharan Africa, the same region already carrying the heaviest burden of childhood malnutrition on the list. The US and UK have both announced large aid cuts over the past year. Those decisions, made in finance ministries thousands of miles away, convert directly into clinic beds, vaccine stock and staff hours in Lagos, Lilongwe and Freetown.
Climate breakdown sits on the same ledger. The commission modelled two paths: a world that holds warming near 1.8°C, and one that reaches 4.4°C. The gap between them is 1.89 billion healthy years. That is the commission's climate number, and it does the same job every Lancet cost-of-inaction figure does — it turns a temperature target into a number a finance ministry has to answer for.
Why a health commission is pricing bioweapons and nuclear winter at all
The Lancet has run global burden-of-disease modelling for three decades, mostly on diseases with known mortality curves: malaria, tuberculosis, heart disease. This is its first attempt to apply the same DALY framework to catastrophic and extinction-level risk — threats with no historical base rate to model from. Christopher Murray, the commission's co-chair and director of the Institute for Health Metrics and Evaluation at the University of Washington, has spent his career building the DALY system into the standard the World Health Organization and World Bank both use to rank disease burden and allocate funding.
Extending that system to nuclear winter and AI-built bioweapons is a methodological wager: that probability-weighted extinction risk belongs on the same table as obesity and alcohol use, ranked by consequence rather than likelihood. The commission's own language concedes the strain — it describes both risks as low-probability, but notes they are the only two on the list capable of ending human life rather than shortening it. The 35-member commission will present the assessment at the World Health Summit in Berlin, and plans to update it annually, adding new threats and data as they emerge.
The catch: a threat only made the list if it had a named fix
Every one of the 17 threats was screened for a second qualifying test, which the commission's own members describe plainly: a high DALY count alone was not enough. Each threat also needed what the commission calls "a realistic route to act." Ibrahim Abubakar, a commission member and professor of infectious disease epidemiology at UCL, framed it as a policymaker's menu, not a horror list — these are the risks where action produces a measurable return, not an exhaustive account of everything that could go wrong.
That screening rule means the list is not a ranking of danger. It is a ranking of danger multiplied by tractability. A threat that is catastrophic but has no identified intervention — say, an asteroid strike with no deflection programme funded — would not clear the bar, regardless of its DALY score. Nuclear conflict and AI misuse cleared it because arms control treaties, export controls on bioweapon precursors, and AI safety regulation are all live policy instruments, however underused. The list, in other words, was built to be actionable first and alarming second. Readers encountering the extinction framing should know it was filtered through a solvability test before it reached the page.
What the commission cannot model: who actually funds the fix
The report costs each threat in lost healthy years, but it does not cost the interventions in dollars. Reaching 1.8°C instead of 4.4°C of warming requires financing commitments no government has fully made. Replacing 1.5 billion DALYs of lost aid requires donor nations to reverse cuts they only just announced. The commission names the benefit of action with precision — 1.89 billion years preserved, 1.5 billion years at risk — but leaves the bill for achieving either figure unwritten. That gap, between a health outcome costed to the year and a funding commitment costed to nothing, is where most global health frameworks stall after the press conference ends.
The open question this commission can't answer yet
The honest gap in the report is its own admission: probability. The commission says nuclear conflict and malicious AI carry extinction risk, but assigns no percentage to either event happening this century. Annual updates may eventually produce a trend line. For now, the claim rests on consequence alone — if it happens, there may be no one left to measure the DALYs. That is a different kind of number than the rest of the report offers, and the commission does not pretend otherwise.
Is this the first time extinction risk has been put in a health framework?
It is the first time the Lancet's DALY methodology — built for diseases — has been extended to nuclear and AI extinction scenarios.
Does the report say how likely an extinction event is by 2100?
No. It states both risks are currently low-probability but does not attach a percentage or timeline to either.
Governments reading this report face a split bill. The climate and aid numbers are owed now, in budget cycles already underway. The nuclear and AI numbers are owed later, if ever — but only if treaties, export controls and safety rules are funded before the risk, not after.