
Over four consecutive days, the United States experienced a series of technological failures: a massive outage of Microsoft services paralyzed the systems of thousands of companies, a technical problem at the Federal Aviation Administration (FAA) grounded aircraft across the country, a software error that affected JPMorgan Chase caused another wave of disruptions, a fault in United Airlines’ system delayed flights again, while an outage of a classified National Security Agency (NSA) network left its employees at home, waiting to return to their offices in Fort Meade.
There is still no evidence that any of these failures was caused by hackers, much less that they are connected to one another. Their unusual timing, however, led many people to suspect a coordinated attack, especially during a period of tensions due to the war between the US and Iran and growing fears of artificial intelligence-assisted hacking.
To allay suspicions and find the truth amid so many failures, a guide has been prepared on how to distinguish a cyberattack from a simple, ordinary system breakdown. The starting point, according to the guide, is the assumption that this is an accident. Matthew Prince, chief executive of the internet infrastructure company Cloudflare, says he has often thought something was a cyberattack and then it turned out not to be.
“At this moment, most of what we are seeing is the usual background noise. Things break, and that is simply the nature of complex systems,” he is quoted as saying. Cybersecurity researcher Doug Madory, who leads internet analysis at Infoblox, also says he tracks up to a dozen network outages a day and that only about “one in a thousand” turns out to be caused by hackers.
“One’s default stance should be that it is not a cyberattack unless there is concrete evidence to the contrary,” he stresses. João Tomé, internet research editor at the network data company IPinfo, notes that most major outages that look like attacks are self-inflicted.
Among common causes, he mentions a failed update, a configuration change, a cut cable or a power outage. In such cases it is often difficult to tell which sites and services are really down. Sometimes the problem arises within a basic service such as a large cloud provider, and the failure spreads to its customers.
In other cases, some services appear to have problems but are working normally, and disruptions appear only when they connect to an entity that is down. That seems to be what happened with that week’s banking outage: several banks appeared to have problems, but all of them were linked to certain interinstitutional transactions involving JPMorgan Chase.
Another similar example is when several telephone operators appear to be having difficulties, while in fact all the failed calls pass through a single network that is experiencing an outage. There are also situations where one or several major services are down and false reports, usually accidental, about other outages add to the confusion.
To understand what is happening, it is advisable to look at verified social media accounts and press release sections on the official websites of major technology companies, which are easily found through search engines. The same approach should be used for government agencies and other institutions.
Another important source is status pages, which major digital platforms provide to show whether different parts of a service are working normally. AWS, Google Cloud Platform and Microsoft Azure offer such pages, as do other services such as Meta’s business products, Google Workspace, telecommunications companies and artificial intelligence companies.
When a failure spreads across many services at once and the point of the problem is not clear, network data that monitor the internet on a large scale help. Amanda Meng, a researcher at the Georgia Institute of Technology who takes part in leading the IODA service for detecting and analyzing internet outages, mentions the service’s data and its accounts on Bluesky and Mastodon, as well as tools such as Cloudflare Radar and Watch the Net. Madory of Infoblox also publishes his findings on network outages on his social media accounts.
When IODA detects an outage, Meng explains, the start and end times, the affected network and any other data are examined, to then be linked to what is reported. The scope, location and timing of an outage help determine the suspected cause: a single network affected by a faulty update, an area cut off due to damage to a fiber-optic cable or, in rarer cases, a cyberattack or even a country deliberately cutting off the internet to its citizens to curb protests.
Although these data provide important information, Meng advises returning to the second step: once the affected service is identified, its owners are usually the best source for following updates. “Usually operators have more information about what is actually happening,” she says.
Most of the time, announcements from companies or government agencies admitting fault confirm that the outage was caused by a technical error. In the rarer cases when an outage comes from a real cyberattack or a security breach, the hackers responsible often take responsibility themselves in one form or another.
Ransomware gangs, perhaps the most common source of dangerous internet outages, publish their attacks on dark web sites, where they list targets for blackmail purposes, sometimes with a countdown timer that pushes victims to pay before the stolen data is published.
In other cases, such as the NotPetya or WannaCry worms, two of the most severe cyberattacks in history, the hackers’ self-spreading code left ransom messages behind on infected computers.
