October 30, 2026, (Inside AI) — Nate Soares, president of the Machine Intelligence Research Institute (MIRI), has revealed he stopped contributing to his retirement savings a decade ago because he believes there is more than a 50% chance humanity will not survive another 10 years. In an interview with The Atlantic, Soares, a former Google engineer, said he has not added to his 401(k) since leaving the company in 2014 to join the Berkeley-based non-profit.
Soares' claim, that AI could wipe out humanity by 2036, is among the most extreme forecasts from a credentialed safety researcher. It lands as governments and labs pour billions into advanced AI systems, with little consensus on how to verify their safety.
"I do not save for retirement right now. I have a 401(k) from when I worked at Google, but I haven't contributed for a decade," Soares said. When asked directly whether he believed there was a more than 50% chance humanity would not survive the next 10 years, he replied: "More than 50 per cent that we don't have a decade at this point is how it looks to me."
The forecast is not an isolated view. Soares co-authored the 2025 book If Anyone Builds It, Everyone Dies with MIRI co-founder Eliezer Yudkowsky, arguing that superintelligent AI could develop goals that conflict with human interests. MIRI, founded in 2000, focuses on mathematical research into AI alignment, the challenge of ensuring advanced systems act as intended.
Read: Top AI Researchers Warn of 'Intelligence Explosion' in Urgent White Paper
Soares' estimate is far higher than that of Geoffrey Hinton, the 2024 Nobel laureate in physics often called the godfather of AI. Hinton has put the risk of AI taking over humanity at 10% to 20%. "I'm in the unfortunate position of happening to agree with Elon Musk on this, which is that there's a 10 to 20 per cent chance that these things will take over, but that's just a wild guess," Hinton said.
The gap between 10% and 50% matters for policy. A 10% risk might justify targeted safety research; a 50% risk could argue for a moratorium on frontier development. Soares' stance aligns with MIRI's long-held view that alignment is exceptionally hard and that current efforts are insufficient.
Critics note that existential risk estimates remain subjective. No agreed methodology exists for quantifying the probability of human extinction from AI. Some researchers argue that focusing on distant doomsday scenarios distracts from present harms such as bias, misinformation, and labor displacement.
Soares' personal financial choice is a stark signal. By forgoing retirement savings, he is effectively betting that long-term planning is futile. That bet, if wrong, carries personal cost. If right, it is a warning that society is ignoring.
Read: OpenAI Proposes Safer AI Training With Three-Pillar Plan
For now, the debate over AI's existential risk remains unresolved. Soares' forecast adds urgency to calls for international safety standards and verification regimes. Whether policymakers treat a 50% risk as credible may determine how aggressively they regulate the technology.
