Platform

How it works The analyst team Trust & audit Integrations Deployment Evidence

Solutions

Defense & suppliers Government Healthcare Financial services MSSPs

Company

About Partners Resources Contact Responsible disclosure
Home/Resources/Article

Analysis

Four Outages, Zero Attacks: Why “Outage or Attack?” Is Now a Core SOC Question

In five days last week, Microsoft 365 went dark, every departing flight in the United States was held on the ground, and customers at the country’s largest banks couldn’t log in to their accounts. None of it was a cyberattack.

18 November 2026

But for hours, in each case, nobody could say that with confidence. That gap — between “something is broken” and “we know whether we’re under attack” — is the real story of the week, and it’s a gap every security team should be thinking about.

Five days, four failures

  • Microsoft 365, Nov 12–13. Outlook, Teams, SharePoint Online and Copilot went down across multiple regions, stalling courts, police departments and government offices. Microsoft traced it to an earlier network capacity upgrade, not an attack.
  • Workday, Nov 13. The HR and payroll platform reported its own service disruption the same day.
  • FAA ground stop, Nov 13. A failure in the telecommunications network that links air traffic control centers triggered a nationwide ground stop and thousands of cancellations. The network crashed when it was rebooted.
  • Major banks, Nov 16. Customers at Chase, Bank of America, Wells Fargo, U.S. Bank, Truist and others reported problems. Chase pointed to a software problem on its side, and because so many transactions touch Chase, the trouble showed up at other banks too.

The question nobody could answer fast

Watch how each event unfolded and the same pattern repeats. First, things break. Then everyone asks: is this an attack?

During the FAA outage, the Homeland Security Secretary said there was no sign of a security threat but couldn’t rule anything out. During the bank outage, posts claiming a coordinated cyberattack on U.S. banks spread on social media within hours, long before anyone had evidence either way.

Inside companies, security teams faced the same question at smaller scale. When Outlook stops working and sign-in errors spike, is that Microsoft, or is it someone in your tenant? When payments to one bank start failing, is that their outage, or fraud?

Get that answer wrong in one direction and your team burns a day chasing a vendor’s bug. Get it wrong in the other and you dismiss a real intrusion as “probably just the outage.”

Why it’s so hard to tell

Outages and attacks look alike in the data. Failed logins, timeouts, error-rate spikes and users locked out are symptoms of both. A wave of failed authentications during a cloud outage can look a lot like a password-spraying campaign.

Alert storms bury the signal. When a major dependency fails, every tool that depends on it starts firing. Analysts face hundreds of alerts at the exact moment they most need to find the few that matter.

Change is the common thread. A capacity upgrade at Microsoft. New technology added to an aging FAA system. A software problem at a bank. Most big outages start with a change, but few security teams can line up “what changed” against “what’s alerting” in real time.

Your blast radius includes your vendors. The bank outage showed how one institution’s problem becomes everyone’s. If you can’t see your third-party dependencies, you can’t tell their failure from your compromise.

What good looks like

The goal isn’t to prevent every outage. It’s to reach a confident answer quickly. Teams that handle weeks like this well tend to:

  1. Check the vendor first. Correlate alerts against provider status pages and known incidents before launching a full investigation.
  2. Separate availability symptoms from attacker behavior. Services being down is availability. New admin accounts, unusual process execution, mail forwarding rules, lateral movement and data leaving the network are attacker behavior. Look for the second set specifically.
  3. Tie alerts to recent changes. Know what was deployed, patched or reconfigured in the last 24 hours.
  4. Document the call. Leadership will ask “are we OK?” within the hour. A written disposition with the evidence behind it is the answer.
  5. Stay alert after the outage. Big, public outages make good cover and good bait. Expect phishing that leans on them, like fake password-reset and “service restored” emails.

Where Intruex fits

This is the problem Intruex was built for. Alerts from your EDR and SIEM route to Intruex, where AI agents triage each one, check it against context, and reach a disposition — escalate or close as benign — with the investigation documented.

In a week like last week, that means the flood of outage-driven alerts gets worked through automatically, while the handful showing real attacker behavior get escalated to your analysts with the evidence attached. Your team spends its time on the alerts that matter, and you have a documented answer when someone asks whether you were attacked.

The takeaway

Outages are going to keep happening. Systems are more connected, change faster and depend on the same handful of providers. The organizations that come through them well won’t be the ones that never go down. They’ll be the ones that can say, quickly and with evidence, “this is an outage, not an attack” — or catch the moment when it’s both.

Want to see how Intruex handles an alert storm? Request a demo.

Sources