Computer scientist Geoffrey Hinton has warned that the world could suffer an artificial intelligence disaster comparable to Chernobyl before the end of next year.
Speaking to The Guardian, the 78-year-old Nobel laureate said an imminent technological catastrophe might finally force the public and global political leaders to take the existential dangers of artificial intelligence seriously.
Hinton explained that he used the 1986 disaster at the Chernobyl nuclear power station in Soviet Ukraine as a symbol of an industrial crisis severe enough to compel immediate political action, rather than forecasting a literal nuclear reactor accident.
On 26 April 1986, an explosion and fire destroyed reactor four at the Chernobyl facility, dispersing massive amounts of radioactive contamination across Europe and forcing the permanent evacuation of the surrounding region.
Threats to power grids and financial networks
Hinton outlined concrete scenarios in which automated software could inflict catastrophic harm without relying on science fiction concepts. In particular, he warned that an attack carried out with the assistance of artificial intelligence could trigger a massive voltage collapse across electrical power grids, or disable the international financial system for an entire week.
He asked whether observers could comprehend the full consequences of such a shutdown, adding that while modern society could eventually recover, the crisis would finally force world leaders to react and pay attention. Hinton remarked that every major world leader would take notice, with the exception of Donald Trump.
The warning reflects a deliberate focus on immediate, tangible vulnerabilities rather than distant hypothetical threats. For several years, Hinton has delivered warnings about artificial intelligence that often adopted an apocalyptic tone.
In 2023, he suggested that a future superintelligent machine might temporarily keep human beings around solely to operate and maintain electrical power stations before ultimately eliminating people altogether. By contrast, his latest prediction requires no sentient software or autonomous machines seeking to replace humanity, focusing instead on practical damage caused by malicious human operators using powerful automated systems.
Decades of pioneering neural network research
Hinton is a British computer scientist and professor emeritus at the University of Toronto. He is widely celebrated across the technology sector as one of the godfathers of artificial intelligence for his foundational contributions to training artificial neural networks.
In 1986, Hinton demonstrated alongside fellow researchers the practical power of backpropagation. The mathematical method enables artificial neural networks to adjust internal connection weights to reduce calculation errors, functioning as an essential engine for training modern machine learning models.
In 2012, Hinton advanced the field further by co-developing AlexNet, a pioneering neural network architecture that achieved unprecedented accuracy in image classification. AlexNet sparked the modern revolution in deep learning, establishing the computational blueprint for contemporary systems that recognise spoken language, generate synthetic images, and answer complex questions in natural text.
Between 2013 and 2023, Hinton continued his cutting-edge artificial intelligence research at Google, rising to the rank of vice president at the multinational company led by chief executive Sundar Pichai. He resigned from Google in May 2023 so that he could speak openly about the hazards of the technology he spent his life building, using subsequent interviews, lectures, and panel discussions as a public megaphone for his concerns.
Nobel recognition and documentary warnings
In 2024, Hinton received global recognition when he shared the Nobel Prize in Physics with American scientist John Hopfield. The Royal Swedish Academy of Sciences honoured the two researchers for foundational discoveries and inventions that enable machine learning through artificial neural networks.
Their groundbreaking work adapted analytical tools from statistical physics to construct computational networks capable of learning, storing, and identifying statistical properties within complex data sets. These discoveries provided the theoretical bedrock for the current boom in artificial intelligence.
Hinton shared his latest warnings ahead of a public screening of a documentary titled AI, Probably Nothing to Worry About. The production drew its ironic title from a prompt response generated by an automated chatbot, and it features Hinton alongside Demis Hassabis, his former Google colleague and the head of the company's DeepMind research division.
During the film, Hinton explained why he continues to speak out publicly against technological dangers. He stated that his professional standing has made him an influential spokesperson for the field, and he felt a personal obligation to spend his entire reputation raising awareness about the issue.
Industry criticism and calls for regulation
Hinton also directed scathing criticism at top Silicon Valley executives who control the world's most prominent artificial intelligence developers. He singled out OpenAI chief executive Sam Altman, stating that he does not trust Altman at all.
Hinton argued that leading technology executives are fully aware of the existential and societal dangers posed by their creations, but choose to ignore them in order to maximise corporate revenues. He stated that he believes Meta chief executive Mark Zuckerberg, Tesla and xAI owner Elon Musk, and Altman understand the risks and do not care at all, preferring instead to keep the money.
To prevent catastrophic failures and mitigate malicious attacks, Hinton outlined specific regulatory measures that governments must enforce. He advocated for mandatory safety testing of the most advanced frontier models, rigorous audits performed by independent evaluators, and common binding rules agreed upon across international borders.
Despite his stark predictions, Hinton insisted that he does not view himself as a catastrophist. He expressed hope that artificial intelligence could still produce a wonderful future defined by radical abundance, although he acknowledged that humanity might well have been better off if the technology had never been developed at all.
