On the afternoon of Friday 13 November 2026, the US Federal Aviation Administration (FAA) stopped every departing flight in the country. The nationwide ground stop lasted a little over four hours, roughly 3,500 aircraft already in the air were brought down safely, and operators cancelled about 12,462 flights that day, according to the FAA. Reuters described it as the longest US shutdown of its kind since the 11 September 2001 attacks.
Within hours, one question dominated social media and search: was it the FAA’s new AI system? Three days before the outage, the agency had switched on an AI-supported traffic-flow tool at all 22 of its en route control centres. The answer officials have given is no. What actually went wrong is, in some ways, a more uncomfortable story about old network infrastructure, and parts of it are still unexplained.
What happened on 13 November?
The FAA says that at about 3:30pm Eastern Time it “experienced a technical problem” and issued a nationwide ground stop “out of an abundance of caution”. Transportation Secretary Sean Duffy told reporters the trouble began at 3:36pm, when control centres saw instability in radar and communications. The stop itself began at around 3:54pm, AVweb reported, and was lifted at 8:15pm.
According to the FAA’s own question-and-answer page, published on 15 November, instability began at 17 of its 22 Air Route Traffic Control Centers and spread into several Terminal Radar Approach Control facilities. The agency stresses that this was “not a full outage”: some radio frequencies and radar feeds stayed reliable, controllers kept tracking aircraft, and medical, law-enforcement and military flights continued.
What caused the FAA ground stop?
The FAA’s explanation is that its legacy equipment became unstable when data began flooding the FAA Telecommunications Infrastructure (FTI) network “at a rate of 1,000 connection attempts every 4 to 5 seconds”. That flood disrupted the system and made aircraft communications and radar feeds unstable.
The agency says its initial investigation points to two potential sources of the excess data: a radar site in Amarillo, Texas, and a radio communications site in El Dorado, Arkansas. L3Harris, the FAA’s network service provider, had been carrying out work at the El Dorado site, including deploying a new router. When the inconsistencies appeared, the FAA says, L3Harris blocked the two locations and the network stabilised. The FAA also disabled a “broadcasting function” on all of the new routers as a precaution, and later on roughly 60 similar routers across the system.
Put simply, the official account is that a piece of network equipment at a remote site appears to have started generating far more traffic than the older parts of the network could cope with, and the instability spread. Duffy summarised it at a White House briefing: “Our switches, our routers actually got overloaded with data and they started to give us inconsistencies in the data from those routers and because of that we shut the system down.”
It is important to be precise about status here. The FAA’s wording is that the initial investigation “indicates” two “potential” locations. It says the FAA and L3Harris “continue to investigate the root cause”. No final report has been published.
Did AI cause the outage?
The suspicion is understandable. The FAA had expanded its Strategic Management of Airspace, Routes and Trajectories tool, known as SMART, to all 22 en route centres three days earlier, the Washington Times reported. SMART merges around 200 data streams, including flight paths, schedules, weather, airport capacity and controller staffing, to forecast demand and recommend early routing or schedule changes.
Every official source so far rejects a link:
- The FAA’s Q&A says: “The two systems are not related. SMART is a data aggregator, and it does not feed information into the air traffic control system.”
- Duffy said SMART makes recommendations to controllers but is not connected to the STARS or ERAM systems that controllers use to track and separate aircraft. “There’s no possible way SMART would have impacted what happened,” he said, according to AVweb.
- Air Space Intelligence, the company behind SMART, said the ground stop was “not based on or a result of the SMART system, any product of ASI or our integration with FAA and ATC”, the Washington Times reported.
- Homeland Security Secretary Markwayne Mullin said there were no “footprints” of an attack and the problem appeared to be “mechanically driven”, AP reported.
So on the evidence published so far, the AI tool is not the cause. What has not been published is the underlying technical detail, such as network logs or a timeline from the El Dorado and Amarillo sites, that would let anyone outside the FAA and its contractor check that account independently.
Was it a cyberattack?
The FAA says that “based on all available evidence, we have no indication that this was a cyberattack or an outside nefarious actor.” Mullin said investigators had found no evidence of a cyberattack, while noting the investigation was ongoing.
Did the FAA’s modernisation work cause it?
This is where the official messaging has been less tidy. AP reported that Mullin said on Friday the FAA was installing “new equipment, new technology” and “it was overwhelming the older system.” The FAA’s later Q&A answers the question “Did your upgrades and modernization efforts cause this?” with “No”, blaming instead “the legacy TDM system”. TDM, or time-division multiplexing, is the older circuit-based telecoms technology that much of the network still relies on.
The two positions are not necessarily contradictory. The FAA draws a distinction between routine equipment replacement, such as swapping a router at a remote site, which it says has been done “for decades” without problems, and its broader, separately funded modernisation programme. It says none of the upgraded towers went down. But for readers, the practical picture is the same either way: new equipment met old infrastructure, and the old infrastructure did not cope.
The FAA has also admitted the timing was unusual. Asked why router work was done in the middle of the day, it said maintenance at major facilities is normally done at night, but that this was a routine replacement at a remote facility. “Based on what happened Friday, we will reevaluate when we do replacement/maintenance work,” it said.
Why does one router matter so much?
This is the part of the story that matters well beyond aviation. Air traffic control in the US depends on a national telecoms backbone that carries radar and radio data between remote sites and control centres. If that backbone treats a sudden flood of connection attempts as legitimate traffic, a fault at one site can degrade service far away. The FAA’s decision to disable the broadcasting function on around 60 routers suggests it sees the configuration of that equipment, not just a single faulty box, as part of the risk.
It is also a familiar pattern from recent technology failures. In the same week, Microsoft blamed its global Microsoft 365 outage on servers becoming unexpectedly unavailable and a recent network configuration change, with load cascading onto the remaining systems. Critical systems increasingly fail not because of exotic attacks or rogue AI, but because of ordinary changes interacting badly with complex, ageing infrastructure.
Duffy has asked Congress for $50 billion in emergency funding to upgrade the air traffic control system, including what he described as routers overloaded with data. “This is an emergency,” he said. “We have old switches, old routers. It’s an old system that went down, went down everywhere.”
What do we know, and what is still unknown?
Established, according to the FAA and officials:
- A nationwide ground stop ran from about 3:54pm to 8:15pm Eastern Time on 13 November.
- Instability began at 17 of 22 en route centres; radar and radio were degraded but not completely lost.
- The FAA attributes it to data flooding its FTI network, with two potential source sites in Texas and Arkansas.
- New-router work by L3Harris was under way at the Arkansas site; blocking the two sites stabilised the network.
- Officials say the SMART AI tool was not connected to the affected systems and there is no evidence of a cyberattack.
Still unknown:
- The confirmed root cause. The FAA and L3Harris say they are still investigating.
- Why the flood of data affected so many facilities so quickly, and why it was not contained automatically.
- Whether a full technical report or logs will be published.
- Whether Congress will approve the $50 billion Duffy has requested.
For now, the most accurate answer to “did AI ground America’s flights?” is that the people responsible for the system say it did not, and the explanation they have given points to network equipment and legacy infrastructure. The fuller answer depends on an investigation that has not yet reported.
Sources: Federal Aviation Administration, “Nationwide Ground Stop: Questions and Answers” (15 November 2026); Reuters; Associated Press; AVweb; The Washington Times; Computer Weekly.
