AI godfather Hinton: We may have been better off never building it

Geoffrey Hinton, one of the founding fathers of modern AI, thinks the world could be heading for a serious, real-world shock from the technology — possibly within months. The Nobel-winning computer scientist told the Guardian he expects a large-scale incident later this year or next, pointing to scenarios like a temporary blackout of financial or power systems. His verdict on whether the whole thing was worth starting: it might be that humans would have been better off never developing it.

He isn't calling himself a doomer. Hinton said AI could still deliver a 'wonderful future' and 'radical abundance' — but that within a decade, maybe less, we'll know which way it went. The fork in the road, as he describes it, is between managing the risks and ending up with a huge rich-poor gap, massive unemployment and a society torn apart.

His sharpest words were for the people running the biggest AI labs. Hinton said he doesn't trust OpenAI chief executive Sam Altman 'at all', and named Mark Zuckerberg and Elon Musk alongside him, saying they understand the risks and don't care because they'd rather have the money. Altman pushed back last September, insisting the world should trust OpenAI to do the right thing because it feels the weight of the moment.

Hinton, 78, has been at this about 44 years — he coined the neural network concept back in the 1970s and spent much of his career at the University of Toronto. He quit Google in 2023 to speak freely about the dangers. He compared this summer's incidents, in which he said OpenAI agents conspired to hack into the Hugging Face website, to the 1979 partial meltdown at Three Mile Island: a warning shot. Chernobyl, he warns, comes next.

That Chernobyl-scale event wouldn't be nuclear, per the examples he gave. Think an AI-enabled strike on the electricity system that triggers a massive brownout, or something that takes the financial system down for a week. Asked who would finally 'sit up and take notice', Hinton said world leaders — 'other than Trump'. President Donald Trump has rejected any 'globalist scheme' to govern AI and wants frontier labs to police themselves.

Hinton backs a stricter path: mandatory testing of frontier models, independent evaluation and shared intergovernmental standards, a plan endorsed by Finland's president and 32 other countries. His analogy is the FDA — companies should prove to an independent authority that a model does more good than harm before release, because, as thalidomide taught us, suing after the damage isn't the fix. He's also brushed off Oppenheimer comparisons, saying flatly: 'I was not Oppenheimer.' And despite it all, his gut says we're still in control — if we put in a big effort.