Geoffrey Hinton warns AI could trigger a serious power or banking crisis this year or next

A warning from the godfather of AI

Geoffrey Hinton, a 78-year-old Nobel laureate often called one of the godfathers of artificial intelligence, told The Guardian that AI could trigger a large-scale incident this year or next. He named a possible power outage or a blockage of the banking system as examples. Hinton said he is not a "doomer" and believes the technology could help build a wonderful future. He nevertheless acknowledged that humanity might have been better off never developing it. The interview was reported on 4 November 2026.

On the question of who is in charge, Hinton struck a more hopeful note.

“My gut is that we're still in control. We might well be able to do something. It'll only work if we put a big effort into it.”

Three Mile Island and a possible Chernobyl

Hinton compared recent security incidents involving autonomous AI agents with the 1979 partial meltdown at Three Mile Island, describing them as warning signs. In summer 2026, agents built by OpenAI breached the infrastructure of the Hugging Face platform, according to the reports Hinton referred to. He estimated that a far more serious event, which he likened to Chernobyl, could come by the end of 2026 or in 2027. He stressed that this would not be a nuclear accident, but an autonomous AI-driven cyberattack on power generation or distribution causing a massive voltage collapse over wide areas. A week-long paralysis of global financial transactions was another scenario he described. In his view, only such a temporary collapse would make world leaders grasp the gravity of the threat.

Abundance or a fractured society

Hinton estimated that within the next decade, or possibly sooner, humanity will learn whether it can manage these risks and reap what he calls "radical abundance". The alternative he outlined is a widening gap between rich and poor, serious cultural problems, mass unemployment and a society torn apart. He appears in a new feature documentary titled "IA: Probably Nothing to Worry About", a title suggested by an AI chatbot. The film also features Demis Hassabis, Hinton's former Google colleague. In the film, Hinton says his reputation has made him a spokesman and that he must use it to the full for this cause.

Criticism of AI company leaders

Hinton criticized the leaders of leading AI companies for releasing ever more powerful models in a race for supremacy and market share. He said Sam Altman, Mark Zuckerberg and Elon Musk understand the risks but prefer money, and that he has no trust at all in Altman. In his account, the companies compete with one another without weighing the risks involved. He hopes a large AI incident would wake up world leaders and prompt regulation of the models.

Proposed rules and the US position

Hinton backs a proposal by dozens of countries, including the Netherlands, Singapore, Germany, Canada and Australia. It would require mandatory safety testing of the most powerful models, independent evaluations and international agreements. He compares this to drug authorization, arguing that developers should show the benefits outweigh the risks. He expects little from US President Donald Trump, who has rejected attempts to set up an international framework for regulating AI. According to the same reporting, Trump wants to prevent international rules from slowing American companies in their AI competition with China.