PDN

Microsoft, banks, aviation: three major outages in five days, and no attack

6 min read

Illustration for the article: Microsoft, banks, aviation: three major outages in five days, and no attack

For the past few weeks, our articles have been about flaws and attacks: the PHP2Shell flaw, the Binance theft, hacked supermarket fridges, booby-trapped Android boxes. Last week, the Internet showed another weakness. In five days, three major outages paralyzed email for hundreds of millions of people, grounded every plane in the United States and locked customers of major banks out of their accounts. None of them was an attack. All three started with an update, a setting or a faulty piece of equipment.

Thursday, November 12: Microsoft 365 stops

On Thursday, November 12, in the early afternoon (Eastern time), Microsoft acknowledged problems accessing Exchange Online mailboxes, first in the Americas, then worldwide. Outlook, Teams, Copilot and the Intune device management tool were affected; OneDrive, SharePoint and the Office apps, to a lesser extent. Service was not fully restored until the morning of Saturday, November 14, according to the BBC.

According to press reports, the outage affected users in more than 70 countries. Microsoft ruled out any cyberattack. Its preliminary explanation, as reported in the press, points to a previously undetected defect in network gateway servers, exposed by a change in how traffic was distributed and by maintenance that had reduced capacity. A fix then reportedly triggered a cascading flood of requests. Microsoft’s final report had not yet been published at the time of writing.

In Quebec and Canada: Santé Québec said no impact on care had been observed, but that appointment booking and the sending of results could be slowed down. The CHUM warned that appointment requests would take longer to process. In Toronto, several major hospitals had difficulties, and Unity Health Toronto declared a “code grey,” the code for IT outages, asking people to call rather than write.

Friday, November 13: the American sky empties

The next day, in mid-afternoon, the FAA (the U.S. civil aviation authority) halted all departures in the United States for just over four hours. Planes already in the air were able to complete their flights. According to press counts, between 8,000 and more than 9,000 flights were cancelled.

The cause, according to U.S. Transportation Secretary Sean Duffy: a newly installed router on the FAA’s telecommunications network, part of the replacement of copper with fibre optics, began sending a thousand connection attempts every four to five seconds, which saturated communications at 17 regional control centres. Here again, no attack.

In Canada, no NAV CANADA system was affected, but departures to the United States, notably at Toronto Pearson, were cancelled.

Monday, November 16: U.S. banks

On Monday, thousands of JPMorgan Chase customers could no longer check their balances or make transactions, online, in the app and even in branches. Downdetector received nearly 5,000 reports for Chase, and more than 800 for Wells Fargo. Customers of other banks saw delays on Zelle transfers exchanged with Chase. Around 6:30 p.m., Chase announced that transactions were once again being processed normally.

Chase attributed the outage to a software bug in a hardware system installed on its own premises, and stated that it was not related to a cyberattack. According to U.S. media, nothing links this outage to those at Microsoft or the FAA.

In Canada, we found no confirmation that any Canadian bank was affected.

No attack, but the same pattern

These three outages have different causes, but they resemble each other:

  • A change that goes wrong. A network upgrade at Microsoft, a new router at the FAA, a software bug at Chase. The major outages of recent years have more often come from an internal change than from a hacker: the CrowdStrike update in July 2024, or the Rogers outage in July 2022, which cut off Interac across Canada.
  • Concentration. When a single provider hosts email for hundreds of millions of people, a mistake on its side becomes a global outage. On October 19, a power cut at a Google Cloud data centre in London had already interrupted services hosted in that region for about six hours.
  • The domino effect. An email outage slows down hospital appointment booking; an outage at one bank delays transfers for customers of other banks; a router blocks flights departing from Canada.

Our view: fraudsters don’t go down

An outage is not an attack, but it opens the door to one. Every major interruption is followed by messages posing as the service that went down: “reactivate your Microsoft account,” “your bank transfer failed, confirm your information,” “your flight was cancelled, request your refund.” A worried, rushed public clicks more easily.

It is the same mechanism we see in the fight against piracy: fraudsters follow the news and slip into every moment of confusion, with fake sites, fake login pages and fake offers. We saw it after the Binance theft, with the scams that followed.

What we recommend

To individuals:

  • after an outage, never log back in from a link received by email or text message: go through the official app or type the service’s address yourself;
  • be wary of any request for a password, verification code or card number “to restore access”;
  • keep another means of payment and your bank’s phone number close at hand.

To businesses and organizations:

  • plan a way to communicate when email goes down (phone, backup messaging), as the hospitals did;
  • during and after an outage, monitor the fake sites and fake pages that impersonate their brand, and have them taken down quickly. That is precisely the kind of impersonation we take down for our clients.

Three outages in five days do not prove that the Internet is collapsing. They are a reminder that it rests on a small number of large systems, and that a single mistake is enough for millions of people to notice at the same time.

Sources

On the same topic

Is your content pirated? We can get it removed.