
In the past week, a Microsoft outage , computer problems at a US bank stopped payments and a grounded flights in the US. Even the US National Security Agency (NSA) reported that its secure intranet NSANet had , and Apple’s online services . The cluster of failures prompted rumours of a coordinated cyberattack by a foreign state or a hacking spree powered by AI.
Speculation about such ideas was so rife that the US secretary of state, Marco Rubio, was forced to that the government had “zero evidence or any reason to believe that any of these isolated or individual instances have anything to do with nefarious actors, with cyber hackers, with criminals, with another nation state, with a malign actor of any kind”.
But why are we seeing so many outages all at once? There are certainly hackers, including state-backed ones who stage coordinated attacks. For instance, a boss at the UK’s intelligence, security and cyber agency – Government Communications Headquarters (GCHQ) – recently told New Scientist that Russia and China are staging “relentless” cyberattacks against the UK using AI. AI has also allowed even non-experts to uncover dramatic bugs in widely used software. But that isn’t what we have seen in recent days.
Advertisement
The US Federal Aviation Administration (FAA) put its air traffic control outage down to “routine maintenance” in which copper communications lines were being . The head of the FAA, Bryan Bedford, echoed Rubio in a press conference and said: “We see zero evidence of anything nefarious.”
Bank outages were put down to a technical issue at JPMorgan Chase. Microsoft blamed a trifecta of problems for its email issues: an old software bug, malfunctioning internal data connections and server capacity. As for the NSA glitch, the spy agency that problems arose “during scheduled routine maintenance”, and didn’t respond to a request for comment. Apple pinned its issues on .
The flurry of outages, then, is partly down to coincidence and partly explained by a chronic lack of investment in resilient infrastructure, says at the University of Manchester, UK. He says that not only is there no indication that these outages are linked, but there is no evidence that they are malicious hacks.
“Things are failing all the time,” says Dresner. “This stuff is really fragile.”
Dresner says there has been an increasing trend – exacerbated by new AI models that can create computer code with unprecedented speed – to prioritise new features and updated interfaces over data validation and reliability.
“We all have what a colleague of mine called the ‘Big Mac mentality’: we want it with all the trimmings and we want it with speed. We’re happy because it’s working – until it fails,” says Dresner. “We wanted agility and we ended up with fragility.”
These sorts of failures aren’t rare, says at New York University. “The honest answer is that it isn’t really all happening now. Outages like these happen all the time. What’s unusual this week is that several big ones landed close together just as everyone started looking for a pattern. As for why it all seems to be happening at once, large outages are far more common than people realise,” he says.
at King’s College London says that the technology sector has been developing profitable and desirable tools, but failing to build and maintain the infrastructure on which those tools rely. He says it is like a car firm selling sports cars when the roads are littered with potholes and dead ends. Spending money on fixing or improving infrastructure brings no immediate profit, so is overlooked.
“This is what we built and this is what we have predicted for years and years and years. It’s coming home to roost,” says Stevens. “If you’re going to build an entire global system on demonstrably insecure networks and systems, then this is what you’re going to get. I don’t think there’s anything more than coincidence in the current set of events, it’s just they seem to have come in a short period of time.”
at the University of Greenwich, UK, says that like the Cybersecurity and Infrastructure Security Agency will certainly have worsened matters overall, but that the real reason for the enormous disruption we see when services go offline is that we all tend to use a very small number of digital tools.
“We are depending increasingly on digital environments which tend to be an oligopoly. Five years ago, that wouldn’t be the case. We would have exactly the same event, but not the same impact,” says Loukas.
In many industries, says Loukas, governments have stepped in and broken up monopolies, but he has seen less progress in the digital world. Until there are many thriving, competing rivals in every sector of technology, we will be vulnerable to enormous outages that bring everything grinding to a halt until staff at one company can find a fix.
Stevens says that AI has led to the discovery of a great number of software bugs, which have required lots of patches. That low-hanging fruit has now been picked and we are seeing a slowdown, he says, but that burst of activity is likely to have been a good thing overall for security.
However, there are huge hurdles between us and a more resilient online world, due to the “perverse economics of cybersecurity”, says Stevens. “If you’re going to test something to oblivion [to make sure it’s secure and resilient], then your competitors have already stolen the market. It is what it is, and I’m quite pessimistic about how to address this. It’s an uphill battle, particularly when the economic incentives point in the opposite direction.”