On October 10, The Lancet published a commission naming 17 catastrophic threats to health through 2100, two of them unlikely but able to cause extinction. The Guardian headlined those two as threatening the “end of humanity by 2100”. The Guardian’s headline has two threats, a verb, threaten, and an object as large as language allows: the end of humanity. What it lacks is any measure. The Lancet’s own release says the likelihood of nuclear conflict and malicious AI remains low. The Guardian story gives no probability figure and first says the risk is low in its sixth paragraph. The headline keeps the scale and drops the odds, and 2100, the last year of the model, reads as a countdown. Set CNN beside it. Same research, under an hour later: AI is one of 17 potentially catastrophic threats that researchers say “require targeted action”. One headline counts, hedges and ends on action. The other picks 2 of 17 and ends on extinction. By the release’s own figures, the biggest avoidable losses come from duller threats: obesity at 23.1 billion DALYs (years of healthy life lost), schooling at 16.1 billion, income inequality at 13.8 billion. Malicious AI sits at 10.4 billion, nuclear conflict at 9.35 billion. The tell is the threat with no number, no base rate and no one who can act. A plain sentence from the same facts: the Commission lists 17 threats, rates two as unlikely but able to cause extinction, and asks governments for threat-ready health systems with durable financing, investment in vaccines, antibiotics and climate technologies, and regular independent tracking. The Guardian puts those recommendations in its second-to-last paragraph. Doom asks nothing of you but dread; the report asks something of governments.
