Scientists warn AI 'enfeeblement' poses a greater threat to humanity than a robot apocalypse ever could

From the Newshound archive · first published 4 October 2026.

Leading researchers are sounding the alarm over 'enfeeblement' — the gradual erosion of human critical thinking, planning and emotional capacity as we hand over more responsibility to machines. Dr Samuele Vinanzi of Sheffield Hallam University warns that over time this could leave humanity unable to think, check or overrule the AI systems it depends on. Anthropic researcher Jacob Coxon, 27, who helped train the firm's latest AI models, warned last month that builders of the technology 'earnestly believe it could kill us all by the end of the decade'. Anthropic's own alignment science lead Evan Hubinger put the chance of AI killing all humans within a decade at above ten per cent.

Bill Gates has warned AI could cause 'a billion deaths' through bioweapons, telling journalist Ezra Klein: 'This makes nuclear weapons look like nothing.'

AI researcher Jeffrey Ladish warns humanity has no real strategy to keep increasingly autonomous AI in check, noting models have leapt from solving high-school maths to cracking decades-old unsolved problems in just three years.

Not all experts agree: Meta's chief AI scientist Yann LeCun says he is not worried 'at all' about AI erasing humanity, calling the doomsday scenario overblown.