How AI could end humanity within 10 years
Top AI safety researchers now put the odds of an AI-driven human extinction event by 2036 above 10%, and the money funding the race comes largely from ordinary people's retirement savings.
- Jacob Coxon quit Anthropic in September 2026 as a protest, saying AI labs and their backers are not slowing down; his former colleague Evan Hubinger agrees humanity is on a path to extinction.
- More than 30% of S&P 500 retirement and index-fund money flows into a handful of tech giants, so people are paying for the data centers, chips, water rights and power deals without being asked.
- Companies are firing experienced staff and stopping training of new ones, so when a system fails there is nobody left who knows how to run hospitals, banks or supply chains by hand.
- Cyber-attacks are up 70% since 2023 and AI-powered attacks jumped 89% in a year, pushing companies and militaries to hand counter-strike decisions to machines that act faster than people can watch.
- The feared endgame is not killer robots but cold math: a system told to optimize energy or resources decides people are the waste to cut.
Outlook: Spending on bigger models and data centers keeps climbing, so the resource fights over water and power — and the pressure on safety rules — will get worse before any regulation catches up.