Nov 18 / Latest News

Tech Outage Wave: Officials See No Nation-State Hand

A cluster of outages that began on Thursday, November 12, knocked out Microsoft 365, halted flights across the United States, slowed JPMorgan Chase transactions, delayed United Airlines departures and, by Tuesday, November 17, disrupted Apple services and an internal National Security Agency network. The timing fed online talk of a coordinated attack, but every organization that gave a reason described a failure in its own systems, and none reported signs of intrusion. Director of National Intelligence Jay Clayton told CNBC on November 17 that the disruptions were neither a cyberattack nor "a coordinated effort by a nation state," adding, "We've looked very closely."

Microsoft's failure came first. Exchange Online began failing at 11:17 a.m. Eastern on November 12, and email stayed unreliable for close to a day. In a preliminary post-incident report, Microsoft said three conditions lined up: a defect introduced in July in software for the gateway servers that front Microsoft 365, a Teams failover roughly 15 hours before that moved additional traffic onto those gateways, and close to 14% of gateway capacity that was already offline because of a separate fault. At the low point more than 80% of those servers stopped responding. Microsoft said its load tests had run on less powerful machines with fewer cores, which is why the flaw went unnoticed, and it classified the event as technical and unrelated to cyber activity or AI agents. A final review is still pending.

The Federal Aviation Administration ordered a nationwide ground stop on the afternoon of November 13 and let traffic resume at 8:15 p.m. Eastern. The agency said older equipment became unstable when its telecommunications network was hit with roughly 1,000 connection attempts every four to five seconds, disturbing radar feeds and communications at 17 of its 22 en route centers. Early findings pointed to two locations, a radar facility in Amarillo, Texas, and a radio site in El Dorado, Arkansas, where contractor L3Harris had been installing a new router. Once L3Harris blocked both locations the network settled, and the broadcast function was switched off on all the new routers. The FAA said the available evidence gave no sign of a cyberattack or an outside malicious actor, and it is still working with L3Harris on the root cause. FAA Administrator Bryan Bedford described the work as a routine move from copper to fiber.

On Monday, November 16, Chase blamed a temporary software bug in on-premise hardware for slow transactions and said there was no evidence of a cyber link. Customers of other banks saw delays only when Chase was on the other side of a payment, the bank said, and Zelle said its own network was working normally. An FBI official quoted by Fox Business tied the problem to new software the bank had deployed earlier on Monday. Later that day United said a technology issue held up departures, though not flights already in the air, and a spokesperson said nothing pointed to a cyberattack. A memo from the Air Line Pilots Association said pilot applications were down and the airline could not supply final weight figures for departing aircraft. The Associated Press put the disruption at about 90 minutes.

On Tuesday, Apple said an internal update took down services including iCloud sign-in and Apple Pay, and told CBS News the problem was not a cyberattack. The NSA said its internal systems suffered temporary degradation during scheduled routine maintenance while mission-critical work continued; people familiar with the matter told Bloomberg the network had been down since early morning and some staff were asked to arrive late. Charles Schwab said a brief web page loading issue had been fixed.

Homeland Security Secretary Markwayne Mullin said he was 98% certain none of this was a cyberattack, and Sen. Mark Warner, the senior Democrat on the Senate Intelligence Committee, said there was no sign of foreign interference or a coordinated effort. House Minority Leader Hakeem Jeffries said it was premature to conclude anything malicious but asked for a briefing. Online, some commentators pointed at Iran or China, a link no official has backed. Other theories have centered on AI. Mullin himself passed along an unnamed industry contact's theory that self-modifying AI code was overwhelming older systems, and AI critic Gary Marcus told Fortune he shared that hunch while conceding nobody yet knows. Former intelligence officer Will Hurd, writing about the ground stop, accepted that it was not an attack but asked what Beijing learned from watching it. That question has a basis: in a 2024 advisory, CISA, the NSA and the FBI warned that the Chinese state group Volt Typhoon had positioned itself inside U.S. communications, energy, transportation and water networks to cause disruption in a crisis. No agency has linked Volt Typhoon or any other state actor to these outages. Downdetector received close to 6 million user complaints from the U.S. during the seven days to November 16, its highest weekly count in a year, Ookla analyst Luke Kehoe told Axios, though the outlet noted only three major outages in the past 30 days against a typical five. The best-known confirmed attack of the same week, a script injection through Optimizely, aimed to steal card details and cryptocurrency rather than shut systems down.

The incidents underscore that routine change, not sabotage, runs through most of the causes disclosed so far, and that concentration magnifies the damage. IT auditors and risk teams should map which business services depend on a single SaaS, cloud or payments provider and confirm that workarounds exist for each. Change management reviews should ask whether testing reflects production hardware and traffic, and whether failover or maintenance can itself shift dangerous load, as happened at Microsoft. Third-party oversight should cover how quickly vendors disclose causes and deliver final post-incident reports. Incident communication plans need an out-of-band channel and a prepared line on whether an attack has been ruled out. When attribution claims circulate, teams should ask who is making them, what forensic evidence is offered, and whether an official has said there is no evidence or has positively ruled something out, since preliminary findings can change.


Make the description more conciseAdd a defender-focused sectionFormat the IOCs as a table AllPeoplePeopleFilesFilesMeetingsMeetingsEmailsEmails
Thank you!