No Hacker Required: What the FAA Ground Stop Teaches OT About Change

Friday, November 13, about 3:30 p.m. ET. The FAA issues a nationwide ground stop. Departures freeze across the country, and for almost five hours the conversation online goes exactly where you’d expect. Was it a hack? Was it AI?

Mira, I get it. When a national system stumbles, everybody reaches for the attacker first. But the FAA’s own Q&A, posted Sunday, tells a more useful story for those of us who run control systems. It points to a routine piece of work at a remote site, on a shared legacy network, that ended up with a national blast radius.

Same availability problem as an attack. No hacker in the picture, at least on the evidence so far.

What actually happened

The agency’s “legacy equipment became unstable when data began flooding the FAA FTI network, at a rate of 1,000 connection attempts every 4 to 5 seconds.” FTI is the FAA Telecommunications Infrastructure, the backbone that carries communications and radar data. That flood “created instability with aircraft communications and radar feeds.” The FAA blames the legacy TDM (time division multiplexing) system for the instability, not the upgraded towers and not SMART, which it says “does not feed information into the air traffic control system.”

How bad did it get? Instability hit 17 of 22 Air Route Traffic Control Centers and spread into several Terminal Radar Approach Controls (TRACONs). But the FAA is clear: “this was not a full outage. Some frequencies and radar feeds remained reliable.”

Now, about the “one router” headline. The FAA doesn’t say that. Its initial investigation points to two potential locations: a radar site in Amarillo, Texas, and a radio communications site in El Dorado, Arkansas. L3Harris, the FAA’s network service provider, “had been doing work at the El Dorado site, including deployment of a new router.” L3Harris blocked both locations and the network stabilized.

Then the FAA disabled the broadcast function on “the originating routers” (plural, amigos) and, as a precaution, on “approximately 60 similar routers in the system.” Transportation Secretary Sean Duffy told ABC it was 69. I’m going with the FAA’s number and flagging the gap.

Root cause? Still open. “The FAA and L3Harris continue to investigate.” Duffy said publicly they haven’t determined whether it was installer error or a faulty router. L3Harris says it’s supporting the investigation and admits no cause.

On cyber, the FAA’s exact line is: “Based on all available evidence, we have no indication that this was a cyberattack or an outside nefarious actor.” Read that carefully. It says no indication. It doesn’t say ruled out, and Duffy left the door open for the FBI.

Traffic resumed at 8:15 p.m. ET. About 3,500 aircraft landed safely, and every passenger arrived safely. Operators canceled about 12,462 flights that day, 9,082 of them by major airlines, though the FAA notes that not all of those were directly caused by the stop.

Why this matters, even if it’s not “cyber”

This part is my analysis, not an FAA finding.

The safety system worked. The ground stop was issued “out of an abundance of caution,” critical life-saving, law enforcement and military flights continued, and everybody got down. Stopping departures is how aviation fails safe.

The availability system is a different story. A job at a remote facility destabilized most of the country’s en-route centers. That’s not really a router story. It’s a shared-network story. When one remote site can flood a network that 17 centers depend on, the network itself is the single point of failure.

And look at the fix. Operators didn’t just block two sites. They turned off broadcast on a whole class of similar devices. That’s a common-mode problem: same function, many similar boxes, one network.

The line that jumped out at me is the FAA’s answer on timing. Maintenance at major facilities “is typically conducted at night,” but this was “a routine replacement at a remote facility, which has regularly been done for decades, without any impacts.” Then: “based on what happened Friday, we will reevaluate when we do replacement/maintenance work.”

Familia, that’s a management of change lesson in the FAA’s own words. “We’ve done it for decades” is not a risk assessment.

We’ve seen this movie before, even at the FAA. In January 2023, contract personnel “unintentionally deleted files” while fixing a database sync, and the NOTAM system outage triggered a nationwide ground stop. “No evidence of a cyber-attack” then, either. Different technical failure, same lesson.

The OT parallel: Browns Ferry, 2006

If you work in plants, this will feel familiar. On August 19, 2006, operators at Browns Ferry Unit 3 manually scrammed the reactor after losing both recirculation pumps. According to NRC Information Notice 2007-15, the licensee found the root cause was a variable frequency drive controller malfunction “because of excessive traffic on the plant ICS network.” The NRC noted that this kind of traffic “is sometimes called a broadcast (or data) storm.”

The licensee couldn’t conclusively say whether a failed PLC or excessive traffic from a different source started it. Sound familiar? Two candidate sources, a flood, and a system that couldn’t take it. The corrective actions included disconnecting the drives from the plant network and installing firewalls.

That’s like your kitchen and your office sharing one hallway. Somebody drops a pot in the kitchen, and nobody in the office can hear the phone. Separate hallways, and the noise stays put.

What to do about it

This is guidance OT already has, not rules the FAA was bound by.

  1. Treat “routine” remote work as a real change. OSHA’s process safety rule (29 CFR 1910.119(l)) asks for the technical basis, safety impact, procedure updates, timing, and authorization before a change. Use that checklist mindset even when you’re not a PSM site.
  2. Test and approve before deploying. NIST SP 800-82r3 says to “document, test, and approve changes before deploying them to the OT environment,” and notes that many OT changes happen “only during scheduled maintenance downtimes.” Timing is a control, por favor.
  3. Know your baseline. Configuration management, per NIST, “can detect improper configurations before they negatively impact performance, safety, or security.” If a new device behaves differently from the baseline, you want to know before it reaches the backbone.
  4. Contain broadcast and flood traffic at the edge. Rate limit, filter, and disable broadcast functions you don’t need. A remote site shouldn’t be able to flood your core.
  5. Segment so one site can’t take down many. Browns Ferry’s fix was separation and firewalls. Ask what your shared network lets one remote node reach.
  6. Test redundancy for common mode. The DOT inspector general described a 2025 Newark-area outage where “both primary and redundant communication infrastructure” failed. If your backup shares the same weakness, it’s not a backup.
  7. Hold vendors and contractors to the same change process.

Bottom line

  • A non-malicious change can cause the same outage as an attacker. Plan for both.
  • “No indication of a cyberattack” isn’t a reason to skip management of change. It’s the reason you need it.
  • Blast radius is a design choice. Segment and contain so a remote site stays a remote problem.

Stay sharp out there.

Javier

Sources

  1. https://www.faa.gov/newsroom/nationwide-ground-stop-questions-and-answers
  2. https://www.l3harris.com/newsroom/press-release/2026/11/l3harris-statement-nov-13-ground-stop
  3. https://abcnews.com/Politics/transportation-secretary-duffy-atc-routers-ability-broadcast-data/story?id=138155545
  4. https://www.faa.gov/newsroom/faa-notam-statement
  5. https://web.archive.org/web/20240422100225/https://www.nrc.gov/docs/ML0710/ML071010303.pdf
  6. https://nvlpubs.nist.gov/nistpubs/SpecialPublications/NIST.SP.800-82r3.pdf
  7. https://www.osha.gov/laws-regs/regulations/standardnumber/1910/1910.119
  8. https://www.oig.dot.gov/library-item/46879

Leave a Reply