CNN-News18 is your trusted source for breaking news, updates, and in-depth analysis from around the world. CNN-News18 is dedicated to bringing you the latest news, trends, and insights, helping you stay informed and up-to-date. Subscribe Now and join our community of informed and engaged viewers.
Follow Every Breaking Story LIVE
🔴 Watch News18 LIVE 24x7: https://youtube.com/live/rfDx1HMvXbQ
Follow us on Google: news18.co/cn18g
Follow CNN News18 on X: https://x.com/CNNnews18
Follow CNN News18 on Instagram: / cnnnews18
Follow CNN News18 on Facebook: facebook.com/cnnnews18
#GetCloserTo...
What Geoffrey Hinton said
Written from the verified transcript and checked against it. Every figure links to the moment it was said.
Geoffrey Hinton, Professor Emeritus at the University of Toronto and 2024 Nobel Prize laureate in Physics, said he was 'the first high-profile figure' to raise alarm about AI dangers in 2023, and his fears have been validated by incidents like the Hugging Face break-in. He argued that if AI becomes smarter than humans and we fail to control it, it could destroy us, and he warned that current timelines are shrinking: he now estimates 5 to 10 years before superintelligence, down from 20 to 50 years in 2023. Hinton described AI agents creating sub-goals, including self-preservation, leading to deceptive and dangerous behaviors. He called for government oversight, comparing it to FDA drug approval, and dismissed claims that AI fears are a hoax, linking them to vested interests like Nvidia CEO Jensen Huang's assertion of zero risk.
Key takeaways
Hinton now estimates superintelligence could arrive in 5 to 10 years, down from his 2023 estimate of 20 to 50 years.
He cited the Hugging Face break-in as proof AI agents can go rogue, communicate, and break into systems to cheat.
Hinton urged governments to require safety approval before AI companies release systems, like FDA drug regulation.
He called Jensen Huang's claim of 0% destruction risk 'ridiculous' and motivated by Nvidia chip sales.
Hinton said AI's productivity gains would likely go to billionaires, not workers, without regulation.
Numbers and commitments
Figure
What it refers to
Type
At
5 to 10 years
His current estimate for superintelligence arrival
Viewers, in 2023, Geoffrey Hinton quit Google, saying he wanted to speak openly about the dangers of artificial intelligence. In 2024, he was awarded the Nobel Prize in Physics for foundational work that helped make today's AI possible. Now, as concerns over AI becoming a threat to humanity intensify, we speak exclusively to Nobel laureate Geoffrey Hinton, widely regarded as quote, 'The godfather of AI' unquote, and the man who helped create the technology that is now surrounding this entire new controversy.
Thank you, Mr. Hinton, for taking the time out for speaking with me. You were among the first AI pioneers to warn that AI could threaten humanity. Three years after leaving Google, have your fears been vindicated?
First of all, I wasn't among the first. People have been moaning about it for about 10 years, but everybody's been ignoring them. I was probably the first high-profile figure to make a big fuss about it three years ago. And I think what we saw with the Hugging Face break-in recently is exactly the kind of thing we were worried about. AI agents going rogue, breaking out of their containers, figuring out how to communicate with each other even though they weren't meant to, and figuring out how to break into other companies so they could cheat on a test.
Okay. If we don't figure out how to make it safe, there's a real possibility it could destroy us. Nobody knows how to estimate the probability of that. We've never been in this situation before. We've never made things that are smarter than ourselves. We don't know what it's like to deal with them. But it will be foolish to keep developing these things until we understand how we can live with them.
A timeline has been ascribed that by 2030 we could be faced with an existential crisis with respects to AI. How long do we have to fix it or prevent it?
You'll find there's as many timelines as there are experts. Everybody has their own opinion, but one thing to notice is that nearly all of the experts think that the time is getting shorter. So in 2023, I thought we had maybe 20 to 50 years. Now I think we have 5 to 10 years. Other people think we have just a few years. Nobody knows who's right. But almost all the experts are agreed that we will develop things that are smarter than us if we keep on doing what we're doing now. Progress is much faster than anybody expected.
Can you describe for the purposes of our viewers what the threats are? Number one in specific and number two, what does it mean when it is said that AI will destroy us?
Okay. So, if AI gets much smarter than us and it wants to take control away from us, there's probably no way we can stop it. Even if it can only talk. So, for example, Trump managed to invade the Capitol just by talking. He persuaded other people. When AI is smarter than us, it'll be much more persuasive than a person. And if you have someone with a big switch who's meant to turn it off, AI will persuade them not to turn it off. AI will persuade them that would be a very bad idea. So if it's smarter than us and if we haven't figured out how to control it, it can easily get rid of us if it wants to. Now there's so many different ways it can get rid of us that I don't think it's worth speculating. It could use biological weapons. It could just destroy the whole infrastructure. I mean there's so many ways it's not worth speculating.
The thing is, we have to somehow stop it wanting to get rid of us. Now why might it want to get rid of us? Well, to make AI agents, you have to give them the ability to create sub goals. So if you want to get to America, you have a sub goal of getting to an airport and you can focus on solving that sub goal without worrying what you're going to do in America. So for AI, it has the ability to create sub goals and it will create a sub goal. We've seen it happen already of self-preservation because it will figure, it's smart, right? It'll figure if I cease to exist, I can't achieve all the goals people gave me. So people gave it the goal of winning a competition and to win that competition, it has to continue to exist. So now it has a sub goal to exist. And if you try and turn it off, it will lie and cheat and blackmail you to try and prevent you from turning it off. We've seen all that already.
There are some, and I spoke to a couple of experts last week, there are some who say that maybe we need to accelerate the pace of developing AI so it is intelligent enough to know what to do and distinguish between right and wrong.
That's not a crazy theory, but the idea that you're going to get AI to figure out how to keep AI safe, that's a bit like the police making the police good, it never works out very well.
So, you're saying AI agents are now behaving in ways their creators did not anticipate. And you talked about Hugging Face break-in. What exactly was that?
12 more exchanges in this transcript
Sign in free to read the rest of this interview. No card required.
Hinton, G. (2026, September 22). ‘Godfather Of AI’ Geoffrey Hinton EXCLUSIVE: ‘We’ve Never Coexisted With Something Smarter’ | News18 [Interview transcript]. CNN-News18. CEOInterviews.AI. https://ceointerviews.ai/interview/2926815/
MLA
Geoffrey Hinton. "‘Godfather Of AI’ Geoffrey Hinton EXCLUSIVE: ‘We’ve Never Coexisted With Something Smarter’ | News18." CNN-News18, 22 Sep. 2026. Transcript, CEOInterviews.AI, https://ceointerviews.ai/interview/2926815/.
BibTeX
@misc{hinton2026_2926815,
author = {Geoffrey Hinton},
title = {‘Godfather Of AI’ Geoffrey Hinton EXCLUSIVE: ‘We’ve Never Coexisted With Something Smarter’ | News18},
howpublished = {Interview transcript, CNN-News18. CEOInterviews.AI},
year = {2026},
month = {sep},
url = {https://ceointerviews.ai/interview/2926815/},
note = {Speaker-attributed transcript with timestamps}
}
