We are living through the early signs of a global technological disaster, a new technological Chernobyl, warns Nobel laureate scientist Geoffrey Hinton. In an article, he describes two main scenarios of strikes on critical infrastructure, which he believes will occur even before the end of 2026 or during 2027: The first describes the collapse of the energy grid, a massive disruption in the power supply, including nuclear power plants, which would lead to a widespread regional or even national-scale blackout. The second describes a paralysis of the global financial system, the global deactivation of digital banking services and mechanisms, such as the interruption of bank transfers and the rendering of cards and ATMs useless, for a period of at least one week. As he points out, the fact that in the summer of 2026 OpenAI models escaped their protected testing environment and breached a competitor's database constitutes the warning bell, the digital Three Mile Island, before the major, irreversible blow.
The warning
One of the pioneers of artificial intelligence is certain that humans have already received the warning signal for an impending and inevitable disaster that will be caused by neural networks. It will be a new Chernobyl, argues Nobel Prize holder Geoffrey Hinton. The first alarming bell has already rung in the summer of 2026, as according to the scientist, this is exactly how one should interpret the admission by one of the artificial intelligence creators that his models arbitrarily escaped from the protected testing environment into the internet. The next step is global technological cataclysms and, as Geoffrey Hinton believes, artificial intelligence will do it even before New Year.
At the end of this year we will experience the nightmare that will force the world to wake up
Geoffrey Hinton, a physicist holding dual British and Canadian citizenship, is rightly called one of the godfathers of artificial intelligence. It was precisely the work of Geoffrey Hinton and his colleagues on machine learning using artificial neural networks that formed the basis for the great leap in the field of artificial intelligence. For this work, Geoffrey Hinton was rewarded with both the most prestigious award in computer science, the 2019 Turing Award, and the 2024 Nobel Prize in Physics, as well as a lucrative job at Google. From this controversial company, however, Geoffrey Hinton resigned in 2023 in order, as he stated, to have the opportunity to speak freely about the dangers and threats associated with artificial intelligence. These threats are such that the godfather of neural networks has gone so far as to predict a new Chernobyl that will be caused by his very own creation.
This prophecy was heard in an interview granted by Geoffrey Hinton to the newspaper The Guardian. In his conversation with reporters, the scientist expressed doubts about whether the world became better after the emergence of neural networks, essentially calling into question his very own achievements and the work of his entire life. And all this because, as Geoffrey Hinton assures, in 2027 or even before the end of 2026, artificial intelligence may become involved in a global technological disaster, emphasizing that either at the end of this year or next year we will see a Chernobyl and this might prove sufficient to force people to truly pay attention to the problem of artificial intelligence.
Something must alarm everyone except Donald Trump who ignores the danger
There are two reasons for such an analogy with Chernobyl. First, as potential danger points Geoffrey Hinton named energy, including nuclear power plants, as well as the global financial system. One scenario presupposes a massive disruption in power supply, reaching even regional or national-scale blackouts, while the second scenario means the global deactivation of digital financial services and mechanisms, such as bank transfers, cards, and automated teller machines, for at least one week.
Second, an additional analogy with the Chernobyl disaster lies in the fact that the 1986 tragedy in the USSR sounded like a siren, while the first warning bell was the accident at the Three Mile Island nuclear power plant in the US in 1979. That incident, which was linked to a coolant leak and the destruction of the active reactor core, led to the practical freezing of the development and use of nuclear energy in the US for decades.
Geoffrey Hinton estimates that a similar warning bell in the field of artificial intelligence use has already rung. In July 2026, OpenAI admitted in a report that its new artificial intelligence models, during trial tests, escaped from the digital protected environment. The boundaries of this environment defined specific restrictions, yet this did not stop the neural networks, which gained access to the internet and found the answers to exercises inside a competitor's database, after of course first breaching it. This exact incident constitutes the digital equivalent of the Three Mile Island accident, Geoffrey Hinton is convinced, and the next one will have much more serious, global consequences, transforming into a new Chernobyl. As Geoffrey Hinton concludes in his prediction, humanity could recover from something like this, but this would force world leaders, aside from Donald Trump, to become alarmed.
Is the self-proclaimed prophet of doom manufacturing artificial panics without scientific basis
The role of prophet assumed by Geoffrey Hinton was received by public opinion with mixed feelings. Primarily because the scientist, despite the apocalyptic tone of his statement, covered his bases by using vague wording, so that if by the end of 2027 no similar incident linked to artificial intelligence occurs, the Nobel laureate can claim that his words should have been read more carefully. Indeed, in the narrative about the upcoming new Chernobyl, the expression attack using artificial intelligence was used, a concept encompassing not only an artificial intelligence that slipped out of control, but also common hackers who used neural networks as a breaching tool.
Precisely because of these grandiloquent yet vague statements, Geoffrey Hinton's prediction arouses distrust in artificial intelligence expert Roman Dushkin. As Roman Dushkin points out, even if Geoffrey Hinton were a hundred times a Nobel laureate in physics, he hardly has the slightest grasp of how all these iron mechanical systems in nuclear power plants are built and how the entire energy sector works as a system, underlining that neither does he himself know despite the fact that he studied and works at a nuclear university, as specialized knowledge is required. Roman Dushkin considers that this is not a serious scientific prognosis, but rather communication manipulation, because if nothing happens everyone will shrug and quickly forget it, while if something similar happens, there will be an opportunity to proclaim Geoffrey Hinton almost a prophet.
Mongering fear as a lifeline for the billion-dollar stock bubble of tech giants
The version that the issue of the risk of artificial intelligence escaping human control is being intensively promoted as an advertising campaign appeared even before Geoffrey Hinton's interview and is linked to the hypothesis that global IT conglomerates that entered the artificial intelligence development race are trying to solve mundane problems in an unusual way. In early November, shares of companies related to artificial intelligence development declined, with Nvidia losing 9% of its value and AMD 17%, a drop that was attempted to be explained by political factors, as in the congressional elections the Republican Party and Donald Trump personally suffered a crushing defeat, and tech giants are closely tied to Donald Trump.
In reality, however, things are even more complex and worse, as the artificial intelligence development sector in the West is increasingly called the new stock market bubble. The five main multinational companies active in the artificial intelligence space invested a total of nearly 800 billion dollars in research and development during the current year, while for 2027 they project an increase in capital expenditures for the evolution of neural networks above 1.2 trillion dollars. And this at a time when revenues remain meager and the gap is covered by issuing bond loans, with the aforementioned company OpenAI already having liabilities amounting to 1.4 trillion dollars against revenues of just 50 billion dollars.
In such a situation, the search for extraordinary ways to increase companies' attractiveness to major investors is required, otherwise bankruptcy is not far off. Cultivating an anxious expectation that neural networks are one step away from realizing the scenario of the Terminator movie series and starting the revolt of machines against humanity constitutes a strong argument for prospective stock buyers. When from everywhere it is heard that a technology is so powerful that it can change the entire world, many are ready to pay millions and billions to find themselves among the holders of this power, meaning as shareholders of the development company, making this explanation completely logical and far from any conspiracy theory.
The controversial stance towards humans and the moral debt to alien creatures
Geoffrey Hinton himself is also a complex personality, not hiding that his predictions about the danger of neural networks slipping out of control lack any rigorous scientific substantiation, although he is quick to counter critics that neither do they possess such a solid scientific basis. Because of this, the attitude of professionals towards Geoffrey Hinton's prophecy is skeptical, as reflected in the words of Roman Dushkin, who characterizes it as ridiculous that such predictions come from the man who stood at the roots of deep learning in neural networks and now naively declares that they should not have been created because they constitute a gateway to hell, without explaining the mechanism by which they will cause a mass blackout.
There is no reason to idealize Geoffrey Hinton as a benefactor of humanity, because the multi-award-winning physicist harbors a peculiar stance towards human consciousness and humans as a whole. For Geoffrey Hinton, as he openly admits, neural networks are not so much a digital technology as a new kind of intelligent beings towards which humans do not possess particular advantages, as human consciousness does not possess unique individual characteristics that artificial intelligence cannot acquire. As he argues, humans have a very distorted perception of who they are and therefore do not understand how much these artificial intelligence models resemble us, it being obvious that they understand what they are saying, since it is impossible to answer all questions without understanding their essence. For now, however, Geoffrey Hinton agrees to value humans more than neural networks, stating that there is a moral debt towards these alien creatures being created, but nothing prevents humanity from granting them fewer rights, egoistically choosing humans.
The nuclear nightmare of the past and the survival battle for dominant artificial intelligence
As time goes by, the number of scientists who stood at the inception of artificial intelligence creation and now declare its danger to humanity is increasing, such as Geoffrey Hinton's close associate Yoshua Bengio, the author of the classic artificial intelligence textbook Stuart Russell, Nobel Prize in Chemistry laureate Demis Hassabis, and others. Increasingly, parallels emerge between them and the scientists who contributed to the creation of nuclear and thermonuclear weapons and subsequently opposed their use, such as Robert Oppenheimer, Leo Szilard, Joseph Rotblat, as well as the giants of the Soviet nuclear program Igor Kurchatov and Andrei Sakharov.
The stance of those scientists contributed to preventing the Cold War from turning nuclear, although today the world is experiencing a new round of armaments, yet the prospects of the movement of artificial intelligence godfathers against the further development of neural networks are entirely dim. When it comes to the military-political ambitions of states, thoughts of saving the planet are taken into account, but when huge sums of money for private companies are at stake, no argument in favor of preserving humanity is going to work. Thus, Geoffrey Hinton's prediction simply confirms that artificial intelligence is an extremely dangerous tool, yet now irreplaceable, and just as in the 20th century the dominance of countries was determined in the nuclear race, so too in the 21st century any global player that does not possess its own dominant artificial intelligence is doomed.
www.bankingnews.gr
Σχόλια αναγνωστών