Safety & Security

Microsoft, FAA, Banks and NSA Outages Signal AI Vulnerability Surge

Outages from November 12 to 17, 2026 hit Microsoft, the FAA, major US banks and the NSA's classified network. Nicole Perlroth says mean time to exploit has fallen to negative seven days.

Cybersecurity Expert Warns of ‘Unprecedented Technological Instability’
Cybersecurity Expert Warns of ‘Unprecedented Technological Instability’Diggity Marketing / unsplash
By Rebecca Stone6 min read

Why it matters

  • Outages ran from November 12 to 17, 2026, hitting Microsoft, the FAA, major US banks and the NSA's classified internal network.
  • AI is accelerating discovery of severe and critical vulnerabilities to 3X the number found in the same period last year, Perlroth said.
  • Mean time to exploit has collapsed to negative seven days, Perlroth wrote.
  • Perlroth cited ARTEX, an AI-enabled offensive tool, spreading to third tier nation states, cybercriminals and digital arsonists.
  • Incident response firms are backlogged six months, with retainer firms unable to triage every incident, Perlroth said.

Cybersecurity expert Nicole Perlroth says the outages that hit Microsoft, the FAA, major US banks and the NSA's classified internal network in November 2026 mark "a period of unprecedented technological instability."

The failures ran from November 12 through November 17, 2026, spanning a flagship software vendor, an aviation regulator, major US banks and a defense agency's most sensitive internal network. Reporting on her post describes Perlroth as "one of the world's leading voices on cybersecurity, cyber conflict, and the emerging era of AI-enabled threats." She said she was inundated with questions about whether the pattern continues.

"We need far more analysis of why so much is breaking at once," Perlroth wrote on X. "I don't want to get out over my skis here, and what I am hearing does not contradict the messaging that most of this is the result of screwy updates."

That caveat frames the post as a working diagnosis rather than a finding. Security teams patching at speed need to know whether they face faulty updates or a coordinated campaign before they rewrite release processes. Perlroth is not making that determination yet. She is asking for it.

The stakes sit in the overlap. Aviation, banking and classified defense networks run on much of the same enterprise software, so a bad update travels further and faster than one company's response team can contain. Perlroth points upstream, at the pipeline that feeds those failures: how many flaws get found, how quickly attackers reach them, and how little room remains to fix them before the patch itself becomes the outage.

What does Perlroth think is happening?

Perlroth said a "collective picture" is beginning to form. She broke it into five parts.

  • An explosion in severe and critical vulnerabilities, with AI accelerating discovery to 3X the number of vulnerabilities discovered in the same period last year. Perlroth said this is collectively being referred to as "vulnageddon" or "the vulnpocalypse."
  • Mean time to exploit those vulnerabilities has collapsed to negative seven days. Perlroth said AI allows attackers to find and exploit flaws far faster than software makers have time to fix them.
  • Powerful AI-enabled offensive tools like ARTEX have been thrown into the hands of every third tier nation state, cybercriminal and digital arsonist, which, Perlroth said, logically creates havoc.
  • That capability compresses patching windows and accelerates software updates, raising the likelihood that those fixes break things in unexpected, cascading ways.
  • Pressure to "AI everything" means organizations are rolling out automated AI tooling on top of brittle infrastructure that cannot support it.

How much faster are attackers working?

Perlroth's numbers describe acceleration on two fronts. Discovery has tripled: she said AI is finding severe and critical vulnerabilities at three times the count of the same period last year, and she attached the labels "vulnageddon" and "the vulnpocalypse" to the surge.

Exploitation has outrun remediation. Mean time to exploit, she said, has collapsed to negative seven days, which on her measure puts attackers in front of defenders rather than behind them. Neither figure carries methodology in the post, which fits her stated preference for more analysis before conclusions.

She also named the tool. ARTEX, she said, is an AI-enabled offensive capability now in the hands of "every third tier nation state, cybercriminal and digital arsonist." Perlroth did not spell out its capabilities or its operators. She used its spread to argue that the population able to probe production systems at machine speed now extends well beyond the best-funded state teams.

She tied the two sides into a loop: faster exploitation shrinks the window for a fix, which pushes teams to ship patches faster, which raises the odds that a patch introduces new failures. "This creates enormous pressure to compress patching windows and accelerate software updates," she wrote, "increasing the likelihood these fixes break things in unexpected, cascading ways."

Her fifth point lands on the same seam. "Pressure to just 'AI everything' means we are rolling out automated AI tooling on top of brittle infrastructure that can't support it." The blanket mandate to automate and the brittle substrate beneath it are the two forces Perlroth holds responsible for the wave.

Why does she blame the rush to automate?

Perlroth held the prevailing approach to software development responsible for the fragility she describes. She took that culture to task for helping to create the scenario, and pointed out the failure of seemingly anyone to ask whether any of these steps should be taken.

"My generation grew up in an extraordinary period of technological stability," she wrote. "We worked our way up the software stack, abstracting away layer after layer, until we lost a fundamental understanding of how the underlying systems actually work."

"We were told to 'move fast and break things.' VCS told us to celebrate the idea that 'software would eat the world.' Now we're telling ourselves we can simply 'roll out AI' across everything, as though nothing underneath will buckle."

What did nobody ask?

Perlroth's sharpest charge is procedural. "We never stopped to ask: SHOULD WE?" She expanded: "Should we be digitizing every last nook and cranny of our companies, infrastructure, healthcare, homes, cars, bodies, lives? Especially when the complexity, interdependence, and fragility we were introducing overwhelmed our ability to understand/secure/maintain the systems we built?"

Then the security framing: "Nobody every asked: What if, in the name of progress, we were creating the ripest attack surface in human history?"

Who repairs the stack when it breaks?

Perlroth's answer is that fewer people can. "We are now entirely dependent on software systems that almost nobody understands end to end, that increasingly few people can repair, and that attackers can now probe at machine speed."

The second half of that passage is an operations problem. "We are hearing that incident response firms, the 911 for breach victims, are backlogged six months. Even those paid on retainer lack the bandwidth to triage every incident. The cavalry is disappearing."

A six-month queue sets the practical ceiling on how much of the pressure above defenders can absorb. Perlroth's combination — tripled discovery, shorter windows, a thinner response layer — describes a system with less slack at every joint, and enterprises that automate patching further will lean harder on that layer.

What comes next?

Perlroth closed with an analogy she considers unusually precise. "Software Beirut" is an incredibly apt analogy. We are in for a period of cascading failures, whether triggered by attackers, bad updates, or our own accumulated complexity.

That leaves the next test with the root-cause analysis she says is missing, and with the patching cycles that keep shortening while it is done. For the companies now pressing ahead with AI deployment regardless, her argument is about sequencing: automation layered onto infrastructure few engineers can still trace end to end concentrates risk faster than it adds capability.

Her closing line doubles as the next checkpoint. "This, I'm sorry to say, is just the opening chapter."

Source: webpronews.com

Share this article:

More from Rebecca Stone

Rebecca Stone

Show full bio

Correspondent covering companies, deals and strategy in artificial intelligence and machine learning at AI In Context.

802 articles

Related articles

  1. Nine in 10 Australian firms hit by slow patching risks
  2. Not a cyberattack: US intel chief rules out foul play in week of outages
  3. Microsoft outage hit 74 countries, and the company says it wasn't AI
  4. 85% of Security Leaders Report Patching-Delay Incidents, Azul Survey Finds
  5. AI-Patched Systems, Not Hackers, Caused US Bank, Airline and NSA Outages

« Previous articleNext article »