How would AI actually ‘kill all humans’? Here are the top five most likely scenarios | Toby Walsh
TL;DR
From a dangerous new bioweapon to total societal breakdown, is a ‘superintelligent’ AI capable of wiping out humanity? Earlier this month, artificial intelligence researcher Jacob Coxon resigned from Anthropic after just four months. In an announcement on X, he stated: “The people building AI earnestly believe that it could kill us all by the end of the decade.
Nauti's Take
For teams building AI systems, the practical consequence is a concrete risk register: which capabilities could scale harm, which permissions would be critical, and where would shutdowns actually work? The ten-percent estimate is an individual judgment and should guide safety work rather than serve as a reliable forecast.
Start by auditing tool permissions, model access, and human approval points.
Sources
