Yuval Noah Harari on the threat of AI: ‘Fascism isn’t the worst-case scenario. Fascism is still human’

This week, I hosted a live conversation between Yuval Harari, the historian who wrote a blockbuster history of the world (Sapiens), and Maria Ressa, the journalist who won the 2021 Nobel peace prize, on the subject of our AI future dystopia. I’m pretty sceptical of AI doomers but, as I explained on stage, I interviewed Yuval back in 2015 and he correctly diagnosed the threat Silicon Valley posed way back then. And Maria, who has lived the reality of technoauthoritarianism, is an essential voice on where lack of accountability leads. 

It’s fair to say I’ve never done an event quite like it. I think it’s a measure of how disturbed and confused we all are by what’s happening with AI that there was a 900-capacity crowd in London’s Union Chapel, sitting on hard wooden benches on a Monday evening, and my attempts to bring the event to a close were ambushed by people insistently shouting questions from the floor. TLDR? The core message from both Yuval and Maria is that nothing is inevitable. Don’t believe the bros: the future is ours to shape.

This is an edited extract of the conversation at 5x15 presents: Yuval Noah Harari & Maria Ressa on The State of the World, 5 October 2026.

Maria Yuval, how do you describe this moment?

Yuval Potentially the most important moment, not just in human history but in the evolution of life, because we are creating something which is not just non-human, but not even organic. And if it could turn out to be more intelligent and more powerful than us.

Maria People often compare AI to previous technologies…

Yuval Usually in order to downplay its importance, or the danger. We haven't seen anything like it before. It's not like the atom bomb or the printing press, because AIs are agents, not tools. A tool is something in our hands that empowers us. Look at nuclear energy: whether we produce cheap electricity, or unleash death and destruction on millions, is a human decision. AI is something that can make decisions by itself. This is the most important thing to understand.

Maria How much time do we have to act?

Yuval Nobody has any idea. The people who are in the know, who know what is being developed, are really freaking out. Right now, we are still in control. If we make the right decisions, we can keep it like that. But if we make the wrong decisions, within a very short time our future, our world, will be in the hands of something else. That is the moment we are in.

Maria Some whistleblowers have suggested that we cannot control the AI models. 

Yuval There is no example of a less intelligent entity controlling a more intelligent one. We don't know of any such case in history; we don't know of any such case through evolution. Chickens don't control humans. And the gap between AI and humans is likely to be at least as big.

Maria Should AI development be stopped or paused? 

Yuval This is obviously not what we advocate, because AI can do a lot of good in medicine, in transport, in your school. But don't race to superintelligence. You can focus on using existing AIs for beneficial purposes – in medicine, in transportation. The race to superintelligence is different. It is ideological, almost religious. It is unsafe. This is the biggest psychological and social experiment in human history, and no one knows what the consequences will be.

Maria What does AI mean for religion?

Yuval Humans conquered the world with language. It's not a coincidence that the Bible says “In the beginning was the Word”. Now, for the first time, there is something that can understand and generate language better than us. Everything made of words will be taken over by AI. Then the question for religion, and for every person, is: am I made of words, or of something beyond words? If your conception of religion is that ultimately there is a text, and the text controls everything, then AI can take it over. If you want to keep some kind of freedom, to remain human, the key question for religious people, and for everyone, is: what am I beyond the words?

Carole Well, this is cheery. Nothing to worry about! Maria, take us back to 2016: what happened in the Philippines, and what happened to you?

Maria What I learned was how information operations could be used, and how free speech could be used to pound you into silence – by design. Lies spread faster. By 2017 in the Philippines, we saw attacks against women journalists 10 times more than men. In 2026 I was getting an average of 90 hate messages per hour. Two years later I was arrested. This technology is predatory and extractive. The tech companies build models of each of us, clones of each of us. We never allowed them to do this. Without facts, you can't have truth. Without truth, you can’t have trust. Without these three, we don't have a shared reality. Without that shared reality you can't solve any problems – climate change, health. 

But here's the last part. You can't have democracy. I think we have a few months, if you look at the data, before we fall into fascism. 

Yuval I don't think that fascism is the worst-case scenario. Fascism is still human. And we have survived it in different parts of the world several times. The real worst-case scenario is that humans lose control. And once we lose control to something more intelligent than us, we are unlikely to ever get it back.

When you look at history, you see that humans made terrible mistakes. But no matter how catastrophic – no matter how bad the regime, the war, the economic crisis – we were still here, and we could learn and try something else. This time, if we make certain kinds of mistakes, we will not get the opportunity to learn from them.

And I completely agree that the breakdown of reality is one of the signs. We have the most sophisticated information technology in human history. You can talk with somebody in the Philippines – not just talk, but on the telephone, see their face. But we can no longer agree. We can no longer have a conversation. People on the other side, wherever they are, seem increasingly alien. And this is very strange. 

Think about the US in the 1960s. The differences were much harder than they are today – the era of civil rights, the Vietnam war, the cold war, the sexual revolution, the feminist revolution. And people would still talk. Something very fundamental is being lost – the human ability to actually talk with each other. That is the source of our power. We rule the world not because we are physically stronger than elephants – we are not. We rule the world because humans can cooperate in very large numbers. And now we are losing the ability to cooperate.

Carole AI’s relationship with the truth is tenuous – they hallucinate – yet we are often unquestioning of the output? 

Yuval We want to train our AIs to be truthful. But we don't know how. The truth has certain inherent disadvantages compared to fictions and fantasies. The truth is expensive, it is complicated, and often it is painful. Fiction is much cheaper. If you look at the long sweep of the evolution of life on Earth, a lot of people have the misconception that over time organisms become better and better at finding the truth – otherwise they would not survive. They think that survival is a proxy for truth. But what happened in evolution is that, for a lot of organisms, the best way to survive is to deceive and manipulate, including to deceive yourself. Much of life is based on self-deception, and also much of history. It will be the same with AIs. Until now, humans were the most intelligent entities on the planet. But we were also by far the most deluded entities on the planet.

Carole Is our failure to hold the tech companies to account the reason why we are where we are now?

Yuval  The companies tell us completely contradictory things. When it comes to accountability, the story is: it's just a game. If you steal something online, it's not really stealing, it doesn't count – it's just a game. So, on the one hand, all the rules that human societies have developed over thousands of years – don't steal – don't apply online, because it's just a game, it doesn't really matter. And then, when they need to raise money from investors, they say: we are taking over the world. This is the only thing that matters. The digital will be much more powerful than the physical. You should regulate sandwiches, but regulating AI – that's crazy, it's just a game.

Carole What can people do to challenge this AI moment? 

Maria We have three cascading failures. There's a crisis of truth – without information integrity, we don't have a shared reality. A crisis of truth leads to a crisis of order. In the Philippines, the institutions collapsed in six months. I'd argue it happened faster in America. And that leads to the third crisis, the crisis of accountability. 

Yuval Be part of human communities that are based on real trust between humans. If you’re not part of one, join one. Trust doesn't disappear – it migrates to the algorithms. People tell you “I don't believe journalists, the newspapers, the old media, they're lying”. But the same person has tremendous trust in the algorithms that control social media. They believe the conspiracy theories about the human journalists, and they have total trust in the algorithms. They are totally naive about the machines.

Carole I went to see Sam Altman back in 2015, and I remember him saying: "The problem is, technology is digital and government is analogue. It's just too slow." And one of the things you talk about is that we're actually on a different time-speed now. I mean, is there time – this is my question – to affect what's facing us?

Yuval There is still time. We don't know how long, but there is still time. Time for what? Time for politics. Politics is about making choices between different options. When they tell you there are no different options, this is inevitable, this is deterministic, what they are saying is there is no politics, there is no time for politics. And this is still not true. It could be true in five years, but it's still not true today.

And I look at something like the US midterm elections, and there are very encouraging signs that, despite the best efforts of the companies, AI is becoming an important issue. And politicians are forced to think about it and to take positions. 

We still have time to explain what is happening, give the public a menu of options, and then the public, at least in a democracy, can choose. 

But I would just offer one word of caution here. In the situation we're in, in 2026, simply refusing – refusing to share data, refusing to build data centres – at this point will not be enough. Because 75% of the compute power is in one place, and in the EU it's just 5%. You cannot regulate something if you're not in the game. The Zulu empire could not regulate the industrial revolution. They could pass wonderful laws about regulating industry.

Carole And we're the Zulus here, right?

Yuval Yes.

Carole There's huge amounts of money going into these companies. And there are a lot of people now saying it's a bubble, it's going to collapse. It's going to crash the global economy.

Yuval Humanity has survived a lot of economic crises. It will survive this one. But, if the predictions in the IPOs that raise these trillions of dollars are correct, and they are going to develop superintelligent AI, and they don't know how to control it, this is something that humanity might not be able to recover from. So a global depression could be the best-case scenario.

Carole Just give them one sentence, one word that people can go out into the evening with. What's the one takeaway? Maria?

Maria Act.

Carole Yuval?

Yuval And trust the humans more than you trust the AIs.

Maria Ressa’s The Fight Back: A Citizen’s Guide to Standing Up to Dictators (WH Allen) is published on 15 October (pre-order here at the Nerve Bookshop)Upgrade to membership to fund the Nerve