The Microsoft outage wasn't quite the disaster movie we'd feared, but next time it could be so different: GUY ADAMS

In the Hollywood disaster movie How It Ends, a mysterious computer glitch knocks out all electricity and communication networks.

Flights are grounded, financial markets seize up, and panicking shoppers unable to use their credit cards start looting petrol stations and supermarkets.

Within a few short hours, major cities have been reduced to rubble-strewn hellscapes controlled by rival gangs of armed motorcycle bandits, one of which ends up shooting and killing our tragic protagonist, played by Oscar-winning actor Forest Whitaker.

That Friday's Microsoft crash appears to have had comparatively benign consequences should, of course, be seen as a blessing.

In homes and offices across Britain, many of us were forced to swap email for WhatsApp and make a few extra phone calls. Power grid bosses were reportedly reduced to using fax machines.

The worst-affected organisations were, needless to say, almost all in the public sector. Some schools, local councils and GP surgeries suffered disruption, while a jury in Manchester had to be sent home because the Ministry of Justice couldn't work out how to transport defendants from prison to court.

There was, in other words, no Mad Max-style apocalypse.

Doomsday 'preppers' who, at the behest of Andy Burnham, have filled their sheds with bottled water and clockwork radios, will have avoided having to prise open a single tin of beans.

Millions of users around the world began reporting Outlook was down on Thursday, in what experts have dubbed the 'most disruptive email-related business outage of all time'

Netflix's How It Ends, starring Theo James (pictures), sees mysterious computer glitch disable power and communications across the United States

Yet the incident dubbed 'the biggest IT outage in history' should not be laughed off. For it offers a deeply worrying insight into the fragility of our age, demonstrating how dangerously reliant modern society now is on a tiny handful of US tech firms.

Thanks to their monopolistic grip on the infrastructure of the online age, a glitch to a single computer system on the far side of the globe can quickly spiral into something that affects us all.

In this instance, a poorly designed line of code in one of Microsoft's regular system updates appears to have caused a breakdown in the email system used by roughly 430million people, in almost every country on Earth.

For most, the result is a minor inconvenience. The big fear, of course, is that the next one will be more serious.

Microsoft's troubles began just after 6pm UK time on Thursday, when users were suddenly locked out of the email service Outlook, AI tool Copilot, and its meeting and messaging platform Teams.

A first wave of reported faults peaked at around 6.16pm, according to the website Downdetector, which chronicles IT failures. Within a few hours, it had logged 140,000 incidents in 74 countries.

Initially, two potentially serious causes emerged. The first was some form of cyber-attack.

In other words hackers, either linked to a hostile state seeking to cause disruption, or a criminal gang, were hoping to extract cash from the tech giant, which had made revenues totalling $282billion (£211billion) last year.

The outage affected airlines, judicial systems and even national spy agencies around the world

 Microsoft tried several solutions to mitigate the problems, including isolating parts of its infrastructure and shifting traffic to alternate routes (Pictured: Microsoft CEO Satya Nadella) 

A second, and perhaps more concerning, scenario, was that an AI system had somehow gone rogue. Experts have recently warned that the hacking abilities of 'bots' may soon surpass even the most robust cybersecurity, potentially creating a massive risk to the internet infrastructure.

Earlier this year, for example, AI agents developed by the Silicon Valley firm OpenAI took it upon themselves to break out from a 'virtual sandbox' – a sort of restricted environment where they were being used to conduct tests – and embark on an international hacking spree.

The possibility that AI systems could turn hostile and compromise an organisation such as Microsoft has since been giving security experts sleepless nights.

Evan Hubinger, a researcher at Anthropic, the company behind the AI tool Claude, said in September that the technology is advancing so rapidly that he fears there is a greater than 10 per cent chance it 'could kill all humans' within the next decade. Thankfully, Microsoft swiftly discounted the possibility of such skulduggery this time, with a senior executive Charles Lamanna, announcing at 4am UK time 'there is no indication this is a result of malicious or cyber activity'.

Instead, the problem appears to have been centred on Exchange Online, the cloud infrastructure that powers mailbox access for Outlook and also supports several other Microsoft 365 products.

To best understand what went wrong, we should perhaps think of the Exchange Online system as a sort of virtual oil tanker.

Rather than having a single engine room, it's powered by several million different 'servers', or individual computer hardware systems. Each is located in a large, air-conditioned facility called a data centre. In the case of Microsoft, its centres are spread across secure locations in roughly 34 different countries.

The overall structure is designed to mitigate the risk of the entire system collapsing due to a single server malfunctioning – to use the oil tanker analogy, there will be plenty of propellers left to power the vessel if one fails.

It also prevents Exchange Online from being wiped out by a fire or disaster at a single facility.

Yet as Thursday night's events demonstrate, it is not failsafe.

In the first few hours of the outage, Microsoft said that Exchange Online had been experiencing an unusually high workload, which it was unable to cope with.

But the company later clarified that, in fact, a substantial number of servers had simultaneously become 'unexpectedly unavailable', which had redirected additional traffic to the remaining servers, causing many of them to break down.

As to what made the servers unavailable in the first place, the firm either doesn't know or isn't currently saying.

But Tony Burke, the Australian home affairs minister, revealed yesterday that Microsoft has been briefing governments, saying the whole thing seems to be related to an 'update that has gone awry'.

Tom Worthington, honorary senior lecturer in the School of Computing at the Australian National University, takes a similar view. He told reporters: 'Rather than a cyber-attack, most likely this was caused by an update to software which contained a bug.'

When they first tried to sort the problem out, Microsoft's small army of engineers – which makes up just under half its 228,000 global workforce – found they could bring at least some of the affected servers back to life via two relatively easy fixes: either turning them off and on again, or adjusting the automated checks used to determine whether they were working properly.

That produced recovery to email and messaging services in certain regions. But it soon emerged that the problem hadn't actually been fixed: a significant number of the servers almost immediately became unresponsive again once normal incoming traffic resumed.

The company therefore began the laborious process of rebuilding servers to a previous software configuration which, presumably, pre-dated the update which caused the initial problem.

After around 20 hours, the issue appears to have been resolved. By last night, emails were pinging their way into Outlook inboxes and Microsoft's share price was back where it had been 24 hours earlier – up around 30 per cent in the past six months.

In other words, it was business as usual. The fear, of course, is that next time could be different.