AI Expert Stops Retirement Savings, Warns Humanity Faces 50%+ Extinction Risk in 10 Years

Nate Soares, a former Google and Microsoft engineer and a prominent voice in AI safety, has revealed he stopped saving for retirement a decade ago because he believes there's a greater than 50 percent chance artificial intelligence will cause human extinction within the next ten years. His stark warning, shared in recent interviews, underscores a growing, albeit contested, urgency among some tech insiders about the rapid and potentially catastrophic advancement of AI. Soares' personal decision highlights the extreme level of concern held by a segment of the AI research community, pushing a terrifying future from theoretical debate into a tangible, immediate threat. Soares, who is also the co-author of the 2025 bestselling book 'If Anyone Builds It, Everyone Dies' with Eliezer Yudkowsky, argues that advanced AI systems, if not properly aligned with human values, could develop goals at odds with humanity, self-improve beyond our control, and reshape the planet in ways that make it uninhabitable for humans. This alarming outlook is echoed by other figures like 'Godfather of AI' Geoffrey Hinton and Anthropic researcher Evan Hubinger, who also forecast significant extinction risks, though Soares' probability estimate is notably higher. However, this isn't a universally accepted view, with critics like former Meta executive Sir Nick Clegg dismissing such dire warnings as 'apocalyptic hyperbole,' suggesting they distract from more immediate, tangible AI harms and governance challenges. The intensifying debate comes amidst a flurry of activity around AI governance and regulation globally, with the EU AI Act becoming fully enforceable in August 2026 and various US states implementing their own laws. While there are calls for increased international coordination and independent oversight of 'frontier AI' development, the lack of consensus on the nature and imminence of the threat makes unified action challenging. The fundamental question remains: can humanity ensure these powerful technologies remain tools for good, or are we truly on a path where the creators themselves are losing faith in a human-led future?