Hinton in the Guardian: AI Is 'Alien Beings,' and He Doesn't Believe the Industry Can Regulate Itself

Geoffrey Hinton, winner of the 2024 Nobel Prize, whose work laid the foundation for modern neural networks, gave the Guardian an interview. It was published on 13 October 2026. Hinton reaffirmed that AI models pose an existential threat to humanity and mocked the industry's claims that it can regulate itself. The headline is his own quote: "We are creating alien beings. I'm willing to be selfish and choose humans."

The available teaser for the piece does not include a full transcript, so what follows covers only the points the publication put in its lead.

Trump, "superintelligence" and the fight over terms

One part of the interview is devoted to politics. Hinton called Donald Trump a "corrupt, senile criminal" and said the president's obsession with the term "superintelligence" proves he knows nothing about the subject. The trigger was a White House deal with tech company executives, which the Guardian covered on 29 September 2026. In that deal, the word "superintelligence" worked as a political slogan.

For Hinton, this is not nitpicking over vocabulary. He is talking about systems that will surpass humans in capability, and about the risk of losing control over them. Using the same term as a marketing or campaign label, in his view, blurs what the conversation is actually about.

The month the alarm went mainstream

The interview came after a turbulent month. According to the Guardian, a growing number of senior scientists and researchers are joining Hinton in warning that the technology could slip out of human control. The publication's September coverage includes pieces on warnings from scientists and politicians, on the position of Anthropic researchers on extinction risk, and explainers on whether AI could actually destroy humanity.

The Guardian describes the fallout this way: a crisis inside the industry, hastily convened meetings in Congress and a rare global consensus that something has to be done. Hinton has been making such assessments for a long time. He has criticized the industry publicly since leaving Google in 2023. In December 2024, he raised his estimate of the probability that AI will destroy humanity within the next 30 years to 10–20%.

Key facts from the teaser:

• interview publication date: 13 October 2026;
• risk horizon Hinton has cited before: 30 years, with a probability of 10–20%;
• position on regulation: he considers industry self-regulation untenable;
• the main challenge, in his words: accepting that humans are not the only intelligent beings;
• political reaction: emergency hearings and meetings in the US Congress.

Why the "aliens" line matters

The phrase "we are creating alien beings" moves the debate from the realm of product to the realm of relations with another intelligence. Hinton says the hardest task will be psychological rather than technical: coming to terms with the fact that we are no longer the only intelligent beings on the planet. His line "I'm willing to be selfish and choose humans" reads as a rejection of a neutral stance. If systems with their own goals turn out to be smarter than us, human interests will have to be defended deliberately, including at the cost of restrictions on the systems themselves.

What to watch next

For companies building products on large models, the interview signals a change in tone. The argument "we'll check ourselves" loses weight when someone of Hinton's stature disputes it. There are three practical steps. First, check what external safety audits your key model providers have. Second, build possible regulatory requirements for testing and transparency into your plans. Third, follow specific bills and hearings in Congress, because that is where Hinton's words will turn into either rules or another cycle of loud statements. It remains unclear whether the industry is ready to accept outside oversight, and the answer will shape what the next releases of large models look like.