«Nobody knows for sure what will happen.» With those words, Nobel laureate Demis Hassabis recently warned about the risk of losing control over AI. It is a warning worth taking seriously: beyond winning the Nobel Prize in Chemistry, Hassabis also runs Google DeepMind.

Until 2021-2022, artificial intelligence was merely a tool embedded in search engines, digital services, and tech devices. Today it is the top priority of every major technology company in the world, and there is no manufacturer or platform that has not built commercial strategies around its development.

You only need to look at the sheer volume of investment in chips, servers, and data centers scattered across every corner of the globe—by private companies and by governments such as China and the United States—to grasp the scale of the sector.

Who is Demis Hassabis, and why his warning cannot be dismissed

It is a development that humanity risks losing control over—and that pushed Demis Hassabis, co-founder and CEO of Google DeepMind and winner of the 2024 Nobel Prize in Chemistry alongside John Jumper, to sound the alarm about the possible consequences of unchecked technological progress.

Hassabis, one of the minds behind projects such as AlphaGo and AlphaFold, believes that artificial general intelligence (AGI)—defined as a system capable of reproducing all the cognitive abilities of the human brain—could arrive within just a few years.

AGI is a system that risks exponentially increasing cybersecurity problems, along with biological and nuclear dangers.

How to put a brake on AI development

For Hassabis, artificial intelligence can be an extraordinary tool, but it must be regulated through strict controls and oversight mechanisms.

The commercial race for AI and its development in the military sphere are delivering enormous progress, but they are also shrinking our understanding of the technology. That is why no one can say with certainty what will happen in the coming years, and why Hassabis is loudly calling for the creation, in the United States, of a body dedicated to evaluating the most advanced AI models.

Such an institution would define the “limits” of models, develop assessments of cybersecurity risks and threats to humanity, and—above all—design tests to flag deceptive behavior by AI.

It is a proposal that, however, risks running into resistance from the labs, which have always been reluctant to disclose data that could put them at a disadvantage in the commercial “battle.”

Hassabis is not the only one raising the alarm

The DeepMind chief’s warning is only the latest call for caution. Earlier, cognitive scientist Geoffrey Hinton acknowledged that we may not be able to control systems more intelligent than ourselves, and even Elon Musk—hardly a figure known for caution—signed a letter calling for at least a six-month pause on training AI systems more powerful than GPT-4.

Uncertainty reigns, then. No one can know whether the AGI Hassabis describes will truly arrive, but this is precisely the moment to develop control mechanisms and limits involving both the public and private sectors. Tomorrow could already be too late.

Editor’s note

This article was originally published in Italian on money.it by Andrea Fabbri on July 22, 2026 as «Nuovo allarme sull’intelligenza artificiale. Il premio Nobel 2024 Hassabis mette in guardia dallo sviluppo indiscriminato dei modelli». It has been translated and adapted for an international audience by the Money.it International desk.