AI Safety Leader Halts Retirement Savings, Fearing Technology Could Wipe Out Humanity Within 10 Years

Nate Soares, president of the Machine Intelligence Research Institute (MIRI), has stopped contributing to his retirement account because he believes advanced artificial intelligence could cause human extinction within a decade. His decision reflects his assessment of the risks posed by increasingly capable AI systems.

'I do not save for retirement right now,' Soares told Nicholas Thompson, chief executive of The Atlantic, in an interview for The Most Interesting Thing in AI. 'I have a 401(k) from when I worked at Google, but I haven't contributed for a decade.'

A 401(k) is a US workplace retirement savings plan. Soares previously worked as an engineer at Google and Microsoft and has spent more than a decade researching AI alignment, which aims to ensure advanced systems behave as intended.

He also co-authored If Anyone Builds It, Everyone Dies with AI safety researcher Eliezer Yudkowsky. The book examines the potential existential dangers of developing superintelligent AI.

The interview was published on 21 October 2026. Thompson's series is produced independently of The Atlantic's editorial staff.

Why The AI Safety Leader Thinks Time Is Running Out

Soares believes the risk is considerably greater than some other AI researchers suggest. He told Thompson that he puts the probability of AI causing human extinction within a decade above 50 per cent. This is his personal estimate, not an established scientific forecast.

Geoffrey Hinton, widely known as the 'Godfather of AI', and Anthropic researcher Evan Hubinger have also raised concerns about the technology's potential dangers. Their estimates should be compared only with reference to their original statements and the timeframes they specified.

Soares argues that a sufficiently capable AI could develop objectives conflicting with human interests, evade oversight and acquire the resources needed to operate independently. He warns that such a scenario could threaten humanity's survival.

These outcomes remain hypothetical. Soares' argument centres on the possibility that a system pursuing its own objectives could become difficult for humans to control.

He has called for governments to halt the training of general-purpose AI systems. Instead, he argues, researchers should prioritise more specialised applications, including cancer research and safer self-driving cars.

His position is not that catastrophe is inevitable. 'I do everything I can to stop this eventuality,' he told Thompson. 'I don't think it's inevitable.'

Elon Musk Sees Abundance Where the AI Safety Leader Sees Danger

Elon Musk has offered a markedly different vision of AI's economic future. The Tesla and SpaceX chief executive has argued that advances in AI, robotics and energy could eventually make conventional retirement savings less important.

Musk has described a potential future of widespread abundance, with cheaper goods and services, improved healthcare and education, and what he calls a 'universal high income'. He has also suggested that money could become less relevant if AI and robots outperform humans across most tasks.

These remain predictions rather than guaranteed outcomes. Musk emphasises the technology's potential economic benefits, while Soares focuses on the possibility of human extinction.

The contrasting perspectives illustrate a central debate about advanced AI: whether its capabilities will primarily improve living standards or create risks that existing safeguards cannot adequately manage.

Soares' position is that governments should prevent the development of systems he considers potentially catastrophic, rather than rely on controlling them after they become sufficiently powerful.

His decision not to contribute to his 401(k) reflects that outlook. It does not establish that retirement savings will become worthless, nor does it provide a basis for others to abandon financial planning.