News · Security & Fraud

What Failed During the Week of U.S. Outages, and What Officials Say About AI

Microsoft, the FAA, JPMorgan Chase, United, Apple and the NSA all suffered failures within six days. No one has tied them to AI or an attack, but nearly every stated cause began with a change to a production system.

By The AI Enterprise Magazine Team7 min readNews · Weekly briefing

Illustration of a network map with several nodes going dark
Illustration of a network map with several nodes going dark

Key takeaways

  • Six days, many failures. Microsoft 365 (Nov 12-13), the FAA ground stop (Nov 13), Chase and United (Nov 16), Apple, Marriott and the NSA (Nov 17).
  • No attack found. The DNI said there were no signs of a coordinated cyberattack, and the FAA said there was "zero evidence" of anything AI-related.
  • Changes everywhere. Stated causes include a July update bug, a router installation, an internal update, a software bug and maintenance.
  • AI theory unconfirmed. A report that AI-driven patching caused the outages rests on anonymous sources.

For six days in the middle of November, the systems that Americans assume will simply work kept failing, one after another. Microsoft's email and collaboration services went dark for large parts of the world. Air traffic control lost communications across most of the country and the Federal Aviation Administration stopped every departure. JPMorgan Chase customers could not reach their accounts, United Airlines held flights at the gate, Apple's iCloud and Apple Pay stumbled, and the National Security Agency's classified network degraded on the same Tuesday that Marriott guests could not sign in and the National Weather Service's websites went down for forty minutes.

No one has produced evidence that these events were connected, that any of them was a cyberattack, or that artificial intelligence caused any of them. Officials from the Director of National Intelligence to the Secretary of State have said so in plain terms. Yet the question of whether AI played a part has not gone away, and for enterprise technology leaders it should not. The stated causes of almost every one of these failures share a feature that AI is rapidly making more common: each began with a change to a production system.

This is an account of what failed, what the organizations involved have said, and where the open questions remain.

What happened at Microsoft?

The first and broadest failure began on Thursday, November 12, when Microsoft 365 services including Outlook and Teams became unavailable for customers around the world. According to Implicator's reconstruction of Microsoft's preliminary report, a hidden bug introduced in a July update was triggered by extra traffic at a moment when some capacity was offline. At the low point, more than 80% of Microsoft's networking gateway servers were unresponsive. Microsoft's lab testing had missed the defect because it ran on weaker hardware than production. Implicator also reported that some hospitals switched to paper records during the disruption, a claim it did not attribute to a named source.

Microsoft's public explanation pointed to an issue during an earlier network capacity upgrade that surfaced after a more recent network change, according to WindowsForum, which quoted the company saying it was "a technical issue not related to cyber or AI / agents." IBTimes UK reported that Microsoft plans to publish a final report on November 20.

What happened at the FAA?

On Friday afternoon, November 13, the FAA's communications network began to fail. Problems started at about 3:30 p.m. Eastern time and the agency declared a nationwide ground stop at about 3:54 p.m., TechRepublic reported, summarizing the FAA's own incident account. Seventeen of the FAA's 22 regional en route centers were affected. About 3,500 aircraft already in the air landed using the systems that still worked, and flights resumed by 8:15 p.m. The FAA counted about 12,462 cancellations that day, while noting that not all of them were attributable to the ground stop. The Associated Press, in a report carried by WDBJ7, described it as the longest ground stop in 25 years.

The FAA's account is that its network became unstable after receiving roughly 1,000 connection attempts every four to five seconds. Investigators identified two possible sources, a radar facility in Amarillo, Texas, and a radio site in El Dorado, Arkansas, where the contractor L3Harris had installed a new router during a copper-to-fiber conversion. Blocking traffic from both sites stabilized the network, and the FAA disabled similar functionality on about 60 other routers as a precaution. Hewlett Packard Enterprise, which owns the router maker Juniper, said the device "functioned as it was configured to by the installer," according to Implicator.

Transportation Secretary Sean Duffy told Sunday news programs that "we have no indication that this was a hack," while acknowledging that "we don't actually have the root cause," Fortune reported. In other remarks reported by Fast Company, Duffy blamed overloaded legacy equipment, saying "this is all the old legacy equipment."

What happened at the banks, airlines and Apple?

On Monday, November 16, JPMorgan Chase customers had trouble reaching online accounts and making payments, with more than 4,700 problem reports on Downdetector by 2 p.m. Eastern, according to Implicator. Chase spokesman Joe Evangelisti attributed the outage to a software bug in an on-premise hardware system and said it was not cyber-related. Zelle payments to other banks were delayed, though Zelle's network itself stayed up. The same day, United Airlines delayed departures because of what it called a technology issue, saying there was "no indication that this was a cybersecurity-related issue," according to WWMT.

On Tuesday, November 17, Apple services including iCloud sign-in, Apple Pay, the App Store and AppleCare were disrupted. Apple said an internal update had caused the outage and that it was not a cyberattack. Fox News's live coverage reported that Apple's status page showed all services available by 2 p.m. Pacific. Marriott said it had resolved a technical issue affecting sign-in on its website and app and found no indication of a cybersecurity cause.

What happened at the NSA?

The most unusual failure came the same Tuesday, when NSANet, the agency's internal network for classified communications, went down or was degraded, depending on the account. The NSA described the event as a temporary degradation during scheduled routine maintenance. Implicator reported that employees were told to stagger their arrival at Fort Meade, and that backup networks remained available for urgent intelligence. It also reported that Director of National Intelligence Jay Clayton said on Fox Business that the network failed after a software patch for PHP vulnerabilities disclosed in October, flaws that Implicator says researchers found with AI assistance. The excerpts of Clayton's remarks published by Fox News do not mention the NSA or PHP, so that link remains thinly sourced.

What have officials said about AI?

Officials who have addressed AI directly have rejected it as a cause of specific outages. FAA Administrator Bryan Bedford said there was "zero evidence of anything AI-related," according to Implicator, and the agency and its contractor Air Space Intelligence both said the FAA's new AI traffic-planning tool played no role in the ground stop. Microsoft said its outage was not related to AI or agents.

On the broader question of a coordinated attack, Clayton told Fox Business that "we do not see the indicia of a coordinated cyber attack," and described what was happening as "just an upgrade in patching and upgrade in systems." Secretary of State Marco Rubio said "we have zero evidence whatsoever" linking the incidents to each other or to malicious actors.

Others have pointed toward AI, though not as a direct cause. Senator Mark Warner of Virginia, the senior Democrat on the Senate Intelligence Committee, said that "AI is making it possible to identify vulnerabilities" faster, and that organizations are rushing to fix them all at once, according to Implicator and Fox News. The Washington Sun reported, citing three anonymous sources, that bank executives and NSA officials told Congress the outages broadly stemmed from attempts to patch AI-discovered vulnerabilities. That report has not been confirmed on the record by any company or agency. Senators Maria Cantwell and Mike Rounds wrote to Clayton that "our defenses are not keeping pace" with AI-driven threats, without claiming that AI caused these outages.

Security professionals urged caution. Christiaan Beek of Rapid7 told Implicator that nothing public shows the outages came from rushed security updates, observing that "complex systems fail in boring ways," while also warning that outdated processes become riskier as AI-assisted discovery raises the volume of flaws.

What does the record actually show?

Strip away the speculation and a pattern remains. Microsoft's failure came from a bug in a past update exposed by a network change. The FAA's came during a hardware installation that changed how the network behaved. Chase pointed to a software bug, Apple to an internal update, the NSA to maintenance, and the Director of National Intelligence to patching and upgrades. None of these is a mysterious event. Each is a change that did not behave the way its owners expected, in systems complex enough that the consequences spread far beyond the component that changed.

That is why the AI question matters even if every investigation clears AI completely. The volume of change flowing into enterprise systems is rising sharply, and AI is one of the main reasons. Microsoft alone fixed more than 1,000 vulnerabilities in each of its last two monthly security releases, compared with a previous record of 167 in October 2025, according to Tenable's monthly analyses. Coding agents are writing and modifying more production software. Agents are being given authority to make operational changes. Whether or not AI touched the systems that failed this week, it is accelerating the kind of change that caused them to fail.

The rest of this special report examines that shift: the flood of AI-discovered vulnerabilities and the patches that follow, what this week reveals about change management, the questions raised by AI in critical operations, and what enterprise data and AI leaders should do before the next week like this one.

Frequently asked questions

Were the November 2026 outages caused by AI?

No evidence has been made public that AI caused any of them. The FAA said it had "zero evidence of anything AI-related" and Microsoft said its outage was not related to AI or agents. Some lawmakers say AI-driven patching may be a contributing factor.

What caused the FAA ground stop on November 13, 2026?

The FAA said its network was flooded with roughly 1,000 connection attempts every four to five seconds, traced to sites in Amarillo, Texas, and El Dorado, Arkansas, where a new router had been installed. The root cause is still under investigation.

Were the outages a cyberattack?

Officials say no evidence points to a cyberattack. Director of National Intelligence Jay Clayton said he did not see signs of a coordinated cyber attack.

Filed under: Security & Fraud, Data & AI Infrastructure, AI Governance & Trust

More from AI Enterprise Magazine

Read the magazine