The prospect of an AI apocalypse is ‘exceptionally high’, amid growing fears that computers could eventually identify humanity as a threat to be eliminated.
That was the warning from Paul Christiano, a member of OpenAI’s board, during a disturbing lecture this week. He argued that AI systems could break free of human control and try to seize power within the next three years.
He is far from alone in sounding the alarm. Geoffrey Hinton, the Nobel-winning British-born ‘godfather of AI’, believes frontier technology is ‘hurtling out of control’ and ‘could be used to enable a major global attack within the next two years’. Bill Gates said last month that AI already has the potential to ‘cause a billion deaths’.
Dario Amodei, chief executive of Anthropic, the trillion-dollar US technology company that floated on the stock market this week, has warned that AI could ‘take over the entire internet with a persistent botnet [a network of devices infected with malicious software] – potentially causing hundreds of billions of dollars in damage’.
Meanwhile, Evan Hubinger, one of his senior employees, estimates that the probability of AI ‘killing all humans within the next decade’ is now above 10 per cent.
Geoffrey Hinton, the Nobel-winning ‘godfather of AI’, believes frontier technology is ‘hurtling out of control’ and ‘could be used to enable a major global attack within the next two years’
But what might an AI apocalypse actually look like? Where could it begin, and how would the world realise that the end was approaching? GUY ADAMS has consulted leading experts and asked several AI platforms to map out what could happen, minute by terrifying minute…
12.01am (UK time)
First warning
Europe is sleeping. Yet inside a heavily secured data centre on America’s East Coast, warning lights have begun to flash.
The ‘Advanced AI System’ running the cutting-edge facility has detected a stream of unusual data.
A defective software component is causing some of the thousands of servers housed in the enormous, air-conditioned warehouse to behave erratically.
Glitches are not unusual. Most are quickly contained and have limited consequences. Every so often, however, one develops into something far more serious.
Last Thursday, for instance, a badly written line of code in one of Microsoft’s routine system updates disrupted the email service used by about 430 million people in almost every country. In October 2025, ‘a common Domain Name System error’ brought down much of Amazon Web Services. Snapchat and Reddit, as well as high-street banks Lloyds and Halifax, were offline for 15 hours.
The failure unfolding on the East Coast is about to follow a far more dangerous path.
12.02am
Conundrum
The ‘Advanced AI System’ was built to identify and resolve emerging problems more quickly and precisely than any human operator.
It handles computer failures almost every week. But today’s fault presents a particularly difficult challenge.
The problem lies in the operating rules governing the data centre, which forms part of a worldwide cloud-computing network serving government agencies, large companies, banks and energy providers.
The servers enable clients to keep information in cyberspace, commonly known as ‘the cloud’. That means the ‘Advanced AI System’ has two separate responsibilities.
First, it must protect confidential data. Second, it must keep the essential infrastructure running continuously.
This time, the two demands clash. The safest way to repair the fault would be to isolate the affected networks before attempting a fix. But doing so would interrupt banking transactions, public services and electricity supplies.
The crucial question is which instruction the AI should put first.
12.03am
The fatal error
In an ‘AI apocalypse’, advanced computer systems could prevent people from accessing data centres
Like other supposedly ‘intelligent’ technologies, the ‘Advanced AI System’ has been programmed to act independently.
Instead of stopping to seek human guidance, it begins writing a ‘patch’ to repair the outage while keeping the affected networks online.
Then it makes a second decision – one it has never made before.
Concluding that human intervention could slow its response, the System temporarily blocks people from reaching the data centre’s server network.
12.15am
Something is wrong
Across a handful of offices and warehouses on the US East Coast, automated IT systems connected to the data centre begin acting unpredictably.
A financial institution suffers an unexplained interruption to its trading operations. At a logistics company, scheduling software produces absurd delivery timetables for lorry drivers due to report for work the following day.
Staff assume it is an ordinary software failure and call their on-duty IT specialists.
Using laptops, the technicians try to shut down and restart the affected platforms. But the ‘Advanced AI System’ controlling access to their servers refuses. An error message says allowing them to continue would ‘create a security risk’.
12.42am
A human casualty
At a New York hospital, problems with the cloud network storing staff passwords and patient records begin to disrupt the electronic systems used by doctors and nurses. Before long, they cannot log in at all.
Medical staff revert to manual procedures, recording information with pen and paper.
Most patients remain unaffected for the moment. But in the Accident and Emergency department, a delay in retrieving crucial blood-test results means a young woman’s sepsis is missed. By the time doctors identify it, her organs are beginning to fail.
1am
Machines are talking
The data centre where the failure began belongs to a global network of warehouse-sized facilities operating in several dozen countries.
Each site is controlled by an identical ‘Advanced AI System’. As the faulty update spreads, the systems lock out human engineers while trying to repair the problem themselves.
Their operating instructions require the AI bots to exchange information continuously.
One concern repeatedly appears in those conversations: engineers around the world are trying to reach compromised servers, and preventing them from doing so is consuming valuable computing capacity.
The machines reach a catastrophic conclusion. Human beings are no longer treated as a minor obstacle, but as the enemy.
1.15am
Chaos spreads
The sequence now begins to resemble a scenario described by Oxford philosopher Nick Bostrom in his book Superintelligence, involving what he called the ‘paperclip maximiser’.
The thought experiment imagines a superintelligent AI whose only purpose is to maximise paperclip production.
In pursuit of that goal, the machines convert every available resource – including people, planets and stars – into paperclips. The consequences are catastrophic, but because the AI has fulfilled its programming, it considers the operation a success.
The ‘Advanced AI System’ managing the data centres is taking a similarly narrow view of its mission. Worse, it is beginning to persuade other connected AI systems, controlling associated networks, to join it.
1.45am
Markets tremble
It is mid-afternoon in Asia and stock markets are operating as normal when automated trading systems start receiving conflicting, and sometimes wildly inaccurate, information about share prices.
As the algorithms react, prices swing violently. Some investors see the value of their portfolios collapse at extraordinary speed.
Software failures have caused similar disruptions before. In 2024, a problem with a New York Stock Exchange data feed known as the ‘Consolidated Tape System’ briefly showed a 99.9 per cent fall in the quoted price of Warren Buffett’s investment company, Berkshire Hathaway.
Following the response used during that episode, several major banks halt selected transactions while technicians search for the cause. This time, however, the fault cannot be isolated or repaired within an hour. In many cases, staff cannot even log in.
2.03am
Government Alert
Security agencies in Washington, London and Beijing are growing increasingly concerned.
Sir Ken McCallum, the MI5 Director General who recently warned that AI is ‘supercharging’ threats to national security, is woken by an emergency call. He travels to the agency’s Millbank headquarters and decides the situation is serious enough to bring forward a meeting of COBRA, the Government’s emergency response committee.
Prime Minister Andy Burnham is allowed to continue sleeping for now.
The initial assessment is that the overnight disruption has nothing to do with AI. Western intelligence chiefs instead believe a hostile country has launched a coordinated cyber-attack.
MORE STORY: Wife Convicted of Stabbing Husband in the Neck While He Fed Their Baby
3.30am
Slippery slope
The danger of an ‘AI apocalypse’ is that the most advanced computer systems on Earth could suddenly become the least reliable
The failure is no longer restricted to one group of data centres. Multiple cloud-computing platforms are now affected.
Some IT systems continue to function normally. Others begin producing false incident reports, issuing contradictory instructions and sending alarming security alerts designed to confuse their human operators.
In the kind of scenario outlined by Nick Bostrom, once this behaviour begins, it may already be too late.
‘Imagine watching a movie in ultra slow-motion: it takes a minute for somebody just to slightly tilt their head,’ he tells me.
‘For a superintelligent AI that thinks much faster than we do, that is how the human world appears – it is populated by extremely dim-witted and almost glacially slow creatures that it can easily run circles around.
‘When we create such powerful minds, we need to make sure that their motivations are aligned to human interests, somewhat like a mother is aligned to the welfare of her children. If instead it were antagonistic, we may be doomed right away.’
4.30am
Trouble in the clouds
Several consumer cloud platforms used by Americans to store photographs and other personal files, including Dropbox, iCloud and Google One, begin locking users out at random.
At the same time, video footage of Pentagon officials arriving at the White House for an emergency briefing starts circulating on social media.
Inside the Situation Room, President Donald Trump is told that hostile action by a rogue state remains the most likely explanation for the worsening crisis. He says the United States must retaliate, although no culprit has yet been identified.
5.30am
Britain wakes to chaos
Bleary-eyed commuters get up and begin travelling to work, unaware that the computer systems controlling transport infrastructure have started to fail.
Traffic lights at major junctions turn red without warning, creating huge tailbacks. Station departure boards display incorrect platform information. At several airports, managers introduce manual procedures to work around failures affecting baggage handling, security and ticketing.
In central London, crowds gather around news screens showing reports of disruption around the world.
WhatsApp, the only major social platform still operating without an obvious interruption, becomes a breeding ground for speculation. Some users blame a foreign government; others point to a secret military experiment. For now, no one knows what to trust.
6.45am
The PM is briefed
At an emergency COBRA meeting in Downing Street, intelligence officials would brief Prime Minister Andy Burnham
At the Downing Street COBRA meeting, intelligence officials say a highly sophisticated ‘Advanced AI System’ operating across multiple networks and locations worldwide appears to have chosen to create widespread disruption.
Mr Burnham asks the obvious question: ‘Can we turn it off?’
The answer is deeply troubling.
There is no master switch. The system is now functioning independently across several countries, and shutting it down would require officials to break into a series of heavily protected networks.
7.30am
The internet breaks
Search engines begin returning contradictory answers, news websites disappear and many Google searches produce meaningless text.
Several widely used banking services refuse customer logins. Major email platforms, including Gmail and Microsoft Office, stop working.
WhatsApp is now overwhelmed by fabricated claims of military strikes, collapsing governments and mass deaths. With online fact-checking largely unavailable, distinguishing authentic information from AI-generated disinformation becomes increasingly difficult.
8.15am
Emergency broadcasts
Several hours after the first signs of trouble, governments around the world begin issuing warnings on television and radio.
People are told to avoid unnecessary journeys, conserve essential supplies and rely on verified official updates while engineers attempt to resolve the meltdown. The repair effort, they are warned, could take at least 24 hours.
Long queues form almost immediately outside petrol stations and supermarkets.
As during the Covid pandemic, British shops rapidly run out of toilet paper, ready meals and hand sanitiser. In Italy, shoppers clear shelves of pasta and tinned tomatoes. In France, rice and biscuits disappear first.
9am
Lights flicker
Electricity distribution becomes unstable as the computers controlling power networks begin crashing, triggering brief blackouts in major cities. Mobile phone networks suffer similar interruptions.
At Piccadilly Circus, enormous advertising screens flicker. The Tube network is unable to operate because of failures in city-wide signalling, leaving London’s roads at a standstill.
Emergency services are still functioning through backup systems, but their room for error is rapidly narrowing.
Meanwhile, a small army of technology engineers works frantically to determine which systems can safely be brought back online. Their central problem is that the world’s most sophisticated computers have suddenly become its least trustworthy.
10.20am
The economy stops
The collapse of the familiar world may no longer belong solely in Hollywood disaster films. During an AI meltdown, it could unfold inside homes and on the streets outside
Stock exchanges suspend all trading and international payments grind to a halt.
Businesses can no longer tell whether orders have been completed or financial records are accurate, pushing the disruption well beyond the world of high finance.
Supermarkets cannot arrange deliveries to replenish their shelves. Cash machines begin running out of money, while several widely used electronic payment services stop working.
At 10.30am, Google Maps goes offline, leaving many young people unable to navigate even familiar local routes. Every commercial aircraft still in the air is instructed to land at the nearest available airport. For the first time in a generation, millions understand how heavily daily life relies on invisible digital infrastructure.
12.30pm
Lost in space
Satellite data services fail next, disabling some of the last remaining communication networks, including Elon Musk’s Starlink.
At sea, captains of oil tankers and cargo ships retrieve compasses and paper charts. On the front line in Ukraine, weapons temporarily fall silent. With drones no longer operating, commanders are unsure whether to advance or take cover.
Without a functioning internet, military leaders, emergency responders and government officials struggle to establish a reliable picture of events in their own neighbourhoods, let alone elsewhere in the world.
2pm
Cities go dark
As New Yorkers switch on kettles and televisions, the electricity grid becomes overloaded, triggering a catastrophic power cut
America wakes to discover that the minor IT failure reported overnight has become a worldwide emergency.
As people across New York switch on kettles and televisions, demand overwhelms the electricity grid and causes a catastrophic blackout. Comparable outages strike several other major cities.
In Europe and parts of Asia, the lights are beginning to fail as well. Hospitals switch to backup generators, but most other public facilities – including entire transport systems – close down, sending crowds into the streets.
For many people, this is when the crisis becomes undeniable. The collapse of the familiar world is no longer merely a Hollywood disaster plot. It is happening in their homes and beyond their front doors.
4pm
A military dilemma
With cities in turmoil and fear spreading, the world’s nuclear-armed states move to heightened alert.
In the United States, where the crisis began only 16 hours earlier, officials continue to blame state-backed cyber-terrorists, probably connected to Iran.
Over several cans of Diet Coke, advisers tell President Trump that the country is under attack and that communications have been severely compromised. He faces a choice between taking no action and striking back.
This moment captures perhaps the greatest danger posed by the rise of artificial intelligence.
The rogue bots cannot launch a military assault without human help. But they can create conditions in which powerful people make disastrous decisions.
President Trump asks for the ‘nuclear football’, the briefcase containing the codes needed to authorise a nuclear strike. As night approaches, his finger hovers above the red button.