Field report SE-28 / J8 · sheet /operations

Operations & RestructuringFiled from the territory

A Week of Cascading Outages Puts AI Security Work Under Scrutiny

Microsoft, the FAA, JPMorgan, Apple, Marriott and the NSA all suffered outages within a week. Officials see no attack — but AI-driven patching may be the common thread.

Grid
SE-28 / J8
Surveyed
November 19, 2026
Reading
4 min read
Correspondent
Sophie Lindqvist
As widespread tech outages pile up, AI tools are in the spotlight again
Fig. SE-28 / J8As widespread tech outages pile up, AI tools are in the spotlight again — AI-generated

Field notes

  1. The FAA suffered the longest nationwide ground stop in 25 years, lasting more than four hours.
  2. Microsoft blamed a 'latent defect' in a network fix first deployed over the summer for its global outage.
  3. AI-detected vulnerabilities in PHP and Java triggered urgent patching that caused service disruptions.
  4. Google reported hackers are using large language models and 'agentic frameworks' to reverse-engineer security patches.
  5. The NSA's classified communications network went offline Tuesday 'during scheduled routine maintenance.'

Six major institutions — Microsoft, the FAA, JPMorgan Chase, Apple, Marriott and the NSA — suffered significant system outages within a single week, and the Trump administration says it sees "zero" indication they are linked or caused by cybercriminals. That official reassurance rests on an unusual suspect: not attackers, but the patching meant to keep them out. Microsoft, United Airlines, JPMorgan, Apple and Marriott all maintain no evidence connects the disruptions to a cyberattack.

What actually went offline?

The failures stacked up across sectors, each traced to routine maintenance or software defects rather than intrusion.

  • Microsoft: a global outage late last week disrupted Microsoft 365 services including Outlook, Teams and the AI assistant Copilot. The company attributed it partly to a "latent defect" in a network fix it had begun deploying over the summer — a fix whose purpose Microsoft did not specify.
  • FAA: a network malfunction triggered the longest nationwide ground stop in 25 years on Friday, grounding flights for more than four hours as carriers cancelled and delayed departures.
  • United Airlines: a separate system outage delayed departing flights on Monday.
  • JPMorgan Chase: a software bug affected internal hardware systems on Monday, delaying transactions and sending jitters through the financial system before resolution.
  • Apple: an internal update temporarily disrupted Apple Music, the App Store and AppleCare on Tuesday.
  • Marriott: a technical issue broke sign-ins on its website and app the same day.
  • NSA: an internal network used to communicate classified information went offline Tuesday "during scheduled routine maintenance."

Microsoft additionally stated its outage did not stem from "AI / agents." The company did not respond to an Associated Press request for further details, and the other firms offered no additional comments when contacted by the AP. What remains unsaid: nobody has explained why so many unrelated systems failed in the same week.

Where does AI enter the picture?

AI has become the connecting thread — not as the cause of the outages, but as the tool whose discoveries are forcing rapid, disruptive patching.

AI recently helped detect sweeping vulnerabilities in PHP, a general-purpose scripting language that powers websites worldwide, prompting urgent update calls — and those fixes produced limited outages and service delays of their own. Jay Clayton, President Donald Trump's director of national intelligence, tied the NSA's failure directly to this work. He told Fox News that patching following last month's PHP alert has caused some systems to "not come back up" as expected. "That is what has happened at the NSA today," he said.

Separately, cybersecurity startup Calif said it used AI to uncover a bug in Java, one of the world's most widely used programming languages and crucial to many large companies. Oracle, which owns Java, released a security update, and U.S. cyber authorities issued an advisory Wednesday urging organizations to install it.

What does this change for security planning?

For executives, the practical question is whether defensive patching now carries operational risk comparable to the attacks it prevents. Experts say yes — and warn the exposure will grow.

"As AI adoption accelerates, we are likely to witness more incidents of this nature," said Raj Rajarajan, director of the Institute for Cyber Security at City St George's, University of London. He noted much of today's digital infrastructure was never designed to handle the "computational demands of large-scale AI services."

The asymmetry cuts against defenders. "AI is absolutely making it way easier for attackers to go and launch cyber attacks. It is not quite as good at creating patches for bugs, like fixing problems, as it is exploiting them," said Justin Cappos, a cybersecurity professor at NYU's Tandon School of Engineering.

Google's cybersecurity report this week underscored the point. Hackers and cyber criminals are "moving beyond basic prompting" toward sophisticated AI techniques, the report found — including large language models and "agentic frameworks" (autonomous AI systems that plan and execute tasks) used to reverse-engineer security patches and exploit the vulnerabilities they reveal.

Is anyone accountable for the cascade?

The outages expose what happens when a single central player in tech, banking, travel or government hits an error — and how little redundancy exists beneath it. Democratic Sen. Mark Warner, co-chair of the Senate's Cybersecurity Caucus, called the week a wake-up call on Tuesday.

"It feels like every day brings another reminder that AI is moving faster than our ability to manage its consequences," Warner said. He also argued the Trump administration has not shown enough interest in confronting these challenges.

Still, Warner offered the same qualified verdict as the companies: "Based on what we know today, there is no good reason" — his fuller formulation — "to believe these disruptions are the unintended consequences of efforts to make our systems safer" rather than anything deliberate. The caveat is the phrase "based on what we know today." Official narratives have shifted before, and no company has yet disclosed the purpose of the summer network fix at the center of Microsoft's failure.

Content creator Jake Smith captured the public mood in an interview: "To see several disruptions this close together, I think to a lot of people it's made them realize how many parts of their lives they've handed over to these systems that we still, at this point, don't fully understand." Until the investigations produce detail, that incomplete understanding remains the operative fact.

via fastcompany.com/rss (Source)

Share this article:

More from Sophie Lindqvist

Sophie Lindqvist

Show full bio

Correspondent covering money, funding and market shifts around corporate strategy and chief-executive leadership at CEO Frontier.

483 articles

Filed under

Adjacent sheets

« Previous article

End of sheetSE-28 / J8 · November 19, 2026