
AINewsIQ illustration. It represents an evidence-first investigation, not a reconstruction of the outage or an established system failure.
Operational update — November 13, 2026, 5:46 p.m. ET
FAA Advisory 120 said the nationwide departure stop remained in effect, with traffic being released in tiers as en-route facilities recovered. The FAA attributed the stop to a communications failure. Homeland Security Secretary Markwayne Mullin said investigators were not ruling out possible causes, including whether the new SMART rollout was involved. Air Space Intelligence, SMART's vendor, said the ground stop was not caused by SMART, an ASI product, or its FAA integration. AINewsIQ has found no evidence tying SMART to the outage.
On Friday afternoon, the Federal Aviation Administration stopped departures across the United States after communications problems affected the facilities that manage aircraft between airports. Three days earlier, the FAA had announced that its AI-supported Strategic Management of Airspace, Routes and Trajectories system—SMART—was expanding to all 22 Air Route Traffic Control Centers.
That timing is important. It is not proof.
The responsible question is not whether an AI system can be blamed quickly. It is whether investigators can show what SMART was doing, which systems actually failed, what dependencies were shared, and how operators recovered. A transparent answer could identify a SMART problem. It could also clear SMART and isolate an unrelated telecommunications failure. Either outcome would strengthen trust more than speculation.
What happened
At about 3:54 p.m. Eastern, the FAA issued a nationwide ground stop for departing flights, describing extraordinary circumstances and a safety need. Transportation Secretary Sean Duffy said a communications issue was affecting all Air Route Traffic Control Centers—the 22 facilities that direct aircraft between airports—and that technicians were investigating.
The stop did not mean aircraft already in the air had suddenly lost control or that the national airspace had become unsafe. A ground stop holds departures while traffic managers stabilize the system. Mullin said officials had no indication of a malicious attack. The underlying cause remained under investigation. Officials later said new equipment and technology overloaded antiquated air traffic control systems, producing intermittent communications and radar gaps. They did not publicly identify SMART as the technology involved. A later DOT briefing was more specific: Duffy said legacy TDM switches and routers were overloaded, while newer IP equipment stayed online. He also said SMART was not plugged into STARS or ERAM, the FAA's terminal-radar and en-route automation systems. On November 14, Duffy further narrowed the account: he said a contractor had been replacing discontinued copper wire with fiber at a remote facility and that the work led to a router overloading the air traffic control network. Officials were investigating contractor error or a faulty router. Bedford said the review would determine whether the router was installed or configured incorrectly or was defective. Duffy said work at small, remote sites was performed during the day because it was not expected to affect the wider system, a practice he said would be reevaluated. Bedford also disclosed that maintenance activity was logged on paper, so investigators had to search physical records to determine who was working where.
FAA Advisory 120, issued at 5:46 p.m. Eastern, said the nationwide stop was still active and traffic would be released in tiers as facilities recovered. That is the latest official operational update AINewsIQ could verify at publication time. The FAA subsequently lifted the nationwide stop Friday evening after roughly five hours. Limited departures resumed, while airlines warned that displaced aircraft and crews could take days to recover. DOT said the airspace reopened at 8:15 p.m. Eastern after the affected systems were stabilized and tested. Nick Daniels, president of the National Air Traffic Controllers Association, told Yahoo News that communications with some aircraft were disrupted and controllers moved flights to working frequencies.
What SMART actually does
The FAA describes SMART as a cloud-based advisory system that centralizes roughly 200 streams of information, including weather, flight paths, traffic flow and controller staffing. On November 10, the FAA said SMART had expanded for use across all 22 centers, so it was operational as an advisory platform—not offline—when the ground stop occurred. A specialist team in Washington analyzes and validates the data, then provides scheduling recommendations to the Air Traffic Control System Command Center in Warrenton, Virginia.
That distinction matters. SMART is intended to support traffic-flow decisions. It does not replace air traffic controllers, and the FAA says people review its recommendations. Describing every nationwide aviation technology problem as an “AI failure” would be inaccurate before the architecture, logs and incident timeline are known.
It would be equally careless to say an advisory system is automatically irrelevant. A system can influence operations without directly controlling aircraft. It can share cloud, network, data or authentication dependencies with other services. Those relationships have to be tested, not assumed in either direction.
Why the timing deserves scrutiny—not a verdict
The FAA announced SMART's nationwide expansion on November 10 after an initial rollout in the National Capital Region. The national ground stop followed on November 13. That sequence creates a legitimate investigative question because deployments, configuration changes and new integrations can expose dependencies that were not visible in a smaller rollout.
But chronology alone does not establish causation. The outage may have originated in telecommunications infrastructure, a separate FAA system, a shared service, a configuration change, an external provider or a combination of failures. Investigators need the change history and system evidence to distinguish among those possibilities.
Air Space Intelligence made its position explicit: the company said the stop was not based on or caused by SMART, any ASI product, or ASI's integration with the FAA and air traffic control. That statement is material and should be part of the record. It is also a vendor statement, not a substitute for an independently documented incident timeline.
The shared-infrastructure question is stronger than an AI-failure claim
FAA architecture documents show why investigators should examine common dependencies before blaming an application. The agency's FTI-2 objectives describe telecommunications services carrying controller voice, surveillance radar and sensor data, weather, flight data and automation traffic. The same document says enterprise messaging services implemented within the broader FAA Telecommunications Infrastructure support the System Wide Information Management program, or SWIM.
Those systems are related without being one undifferentiated network path. The FAA's SWIM guidance says SWIM and digital voice communications are both part of the larger FTI environment, but SWIM infrastructure itself does not carry digital voice. That distinction is important: a shared backbone can create common dependencies while individual services retain separate operational routes.
The 2025 Newark disruptions provide a concrete precedent for asking about telecommunications rather than assuming a physical radar failure. The FAA said STARS radar data for Newark was processed in New York and sent over telecommunications lines to the Philadelphia TRACON. Its response included additional high-bandwidth connections, fiber and a temporary backup system to increase redundancy. An even closer precedent occurred in 2009 during another FTI transition to fiber. A Department of Transportation inspector general review found that contractor network-configuration and procedural errors blocked approximately 75 percent of FTI traffic and delayed restoration for five hours. The review warned that future outages remained possible as new services were added to the fiber network and said FAA oversight of the contractor could have been more effective and proactive. That record makes the present incident a test of whether the earlier change-control, monitoring and contractor-oversight lessons were institutionalized—not merely whether one router was defective.
This evidence does not establish SMART's network path. The FAA says SMART consumes roughly 200 data streams, but the public material reviewed by AINewsIQ does not provide a verified SMART network diagram or show that its rollout changed FTI, SWIM, voice or surveillance routing. Nor has the FAA confirmed that radar surveillance failed nationwide during Friday's stop. The stronger hypothesis to test is whether a common telecommunications dependency failed—and whether the recent rollout touched that dependency—not whether “the AI” malfunctioned. Officials now say the failure produced intermittent communications and radar gaps and that new technology overloaded older systems. That strengthens the shared-dependency inquiry without identifying SMART as the affected technology. Duffy later said SMART was not plugged into STARS or ERAM and was disconnected from the equipment that failed. Read narrowly, that describes architectural separation from the affected automation path; it does not mean SMART itself was offline. A detailed technical incident report has not yet been published. The later contractor-and-router account materially narrows the infrastructure hypothesis away from SMART, but it does not yet provide the network-flow evidence needed to independently establish which traffic overloaded the router or exclude every indirect integration effect. Identifying the overloaded router locates a failure point; it does not explain what generated the reported connection storm or whether the rollout’s additional data demands were modeled against the legacy network. The public record reviewed by AINewsIQ contains no FAA load-test plan, capacity model or results showing how SMART’s nationwide 200-stream workload was tested against legacy TDM equipment, shared data services and upstream producers. That absence is a disclosure gap, not proof that testing did not occur. Inadequate load testing remains a plausible hypothesis—not a verified finding—and the FAA should publish the relevant plan, assumptions and results. What the router finding changes is the balance of the evidence: it supplies a plausible immediate failure mechanism and makes a direct SMART malfunction less likely. It does not fully exonerate SMART or its rollout. ‘Did not go down’ is not the same as ‘did not contribute,’ and ‘disconnected from STARS and ERAM’ is not the same as ‘isolated from every shared network service.’ A system can remain operational while its traffic, integrations or deployment changes add load to a shared downstream dependency. Excluding that possibility requires the network-flow logs, configuration changes, dependency maps and capacity-test results the FAA has not yet made public.
The questions existed before Friday
Concern about the rollout did not begin with the outage. In an October 9 letter, Representative Don Beyer and other lawmakers asked Duffy and FAA Administrator Bryan Bedford to pause SMART while the agency answered questions about safety testing, workforce training, oversight and accountability.
On October 13, 23 Democratic members of the House Transportation and Infrastructure Committee asked the Government Accountability Office to review SMART and the FAA's Flow Management Data and Services platform. Their request covered acquisition, testing, cybersecurity, data governance and workforce readiness.
The National Air Traffic Controllers Association told Nextgov/FCW in October that it had not been involved in SMART's design, testing or implementation. The union also said new technology should complement controllers rather than replace them.
Those concerns do not prove a defect. They show why documentation, operational participation and independent review matter before an incident forces the questions into public view.
Ten questions the FAA should answer
1. Was SMART operational in all 22 centers when the communications failure began? The FAA's November 10 statement says it had expanded SMART for use across all 22 centers; a technical report should still specify which facilities and data paths were active at incident time. 2. Which component first failed, and what evidence identifies it as the initiating event? Duffy now points to a router during copper-to-fiber upgrade work; the supporting logs and technical report remain unpublished. 3. Did SMART continue receiving complete, current data during the outage, or did any inputs become stale or unavailable? 4. Did operators accept, reject or act on a SMART recommendation immediately before or during the event? 5. Did SMART and the failed communications systems share a cloud, network, identity, data or monitoring dependency? 6. What deployments, configuration changes or integration changes occurred in the preceding 72 hours; why was work with possible nationwide impact allowed during daytime traffic; and what load, endurance and failure-mode tests modeled the nationwide 200-stream workload against legacy TDM equipment, shared data services and upstream producers? 7. What did SMART display when its inputs or dependencies degraded? 8. Which fallback procedures worked, and which human actions enabled the tiered recovery? 9. Were complete logs, configuration snapshots and operator actions preserved for independent review? What maintenance information existed only on paper, and how did that affect detection and response? 10. When will the FAA publish a root-cause timeline that separates confirmed facts from unresolved hypotheses and includes the network-flow logs, change records, load-test results and any applicable procurement or ethics-review materials?
These are not accusations. They are the questions that can protect a useful system from being blamed by association—or reveal a problem that needs correction.
What the evidence does and does not show
The timing creates a question. It does not show that SMART initiated the communications failure. The later technology-overload attribution narrows the inquiry, but it still does not establish that SMART caused the overload. DOT's later account points to legacy TDM switches and routers, while Duffy said SMART was not connected to STARS or ERAM. That is a specific official attribution, but the supporting technical incident report remains pending. Duffy's November 14 account further points to a router affected during contractor copper-to-fiber work; officials are still determining whether contractor error or faulty equipment was responsible.
The vendor's denial is relevant. It does not independently prove that every shared dependency or integration was unaffected.
The FAA calls SMART advisory and human-reviewed. That limits the conclusions people should draw about autonomous control, but it does not remove SMART from a serious dependency review.
Officials have not identified a malicious attack. That does not yet amount to a final public root-cause finding.
Investigating SMART is not the same as blaming SMART. A proper investigation should be able to exonerate a system as clearly as it can identify a fault.
AI earns trust through evidence
Critical infrastructure cannot ask the public to choose between blind confidence and reflexive suspicion. Trust comes from architecture boundaries that are understandable, logs that survive an outage, tested degradation modes, named human accountability and a public explanation after the incident. Public explanations now pull in different directions: Mullin said new technology was overwhelming older systems, while Duffy said the new equipment remained online and SMART was disconnected from what failed. Both statements could be true—an upstream service can remain operational while its traffic, integration or configuration overwhelms a downstream dependency—but neither resolves the question without topology, flow logs, configuration history and load-test results. Duffy then named L3Harris and raised contractor “incompetence” or a faulty router before investigators had established which; L3Harris said it was supporting the investigation to understand the conditions that led to the stop. Its statement did not accept responsibility or identify what failed. Naming the maintenance activity is relevant evidence. Assigning responsibility before publishing the supporting evidence is not transparency. Duffy is also not a disinterested narrator of a rollout he championed: the FAA awarded ASI an $875 million SMART and FMDS contract under his leadership, and campaign-finance records show ASI’s PAC contributed $2,500 to the congressional campaign of his son-in-law before the award was announced. That sequence does not prove improper influence, procurement misconduct or a SMART role in the outage. It creates a potential appearance-of-conflict issue that makes independent technical and procurement review more important.
If SMART was unrelated to Friday's outage, the strongest defense is not a slogan. It is evidence showing where SMART sat in the operating chain, what it received, what it recommended, what it could not control and why the actual failure occurred elsewhere.
If SMART or a shared dependency contributed, transparency still serves the system. It allows the FAA and ASI to correct the problem, retest the boundaries and explain what changed before the next nationwide event.
Trust in AI is not rewarded by doublespeak. It grows when officials, vendors and contractors disclose the same underlying evidence, acknowledge uncertainty and allow independent reviewers to test their claims. That is how AI earns trust in infrastructure: not by being protected from questions, but by being designed and operated so the questions can be answered.
Update log
- 3:54 p.m. ET: The FAA issued a nationwide departure ground stop for safety reasons.
- 5:34 p.m. ET: Air Space Intelligence said SMART and its FAA integration did not cause the stop.
- 5:46 p.m. ET: FAA Advisory 120 said the stop remained active and traffic would be released in tiers as en-route facilities recovered. The FAA later lifted the stop Friday evening after roughly five hours. DOT said the airspace reopened at 8:15 p.m. Eastern.
- 7:10 p.m. ET — Architecture context: FAA material confirms that voice, surveillance, weather, flight-data, automation and SWIM services can depend on the broader FTI environment. It does not establish SMART's network path or a causal connection to the outage.
- Cause: The FAA described a communications failure; the root cause remained under investigation at publication. Officials later attributed the disruption broadly to new technology overloading antiquated systems and causing intermittent communications and radar gaps; the initiating component and dependency chain were not publicly identified. A later DOT briefing specified that legacy TDM switches and routers overloaded while newer IP equipment stayed online, superseding the earlier broad description.
- SMART: Officials said investigators would examine possible causes. No connection to SMART had been established. No public evidence reviewed by AINewsIQ establishes that SMART caused or contributed to the outage. Duffy subsequently said SMART was not plugged into STARS or ERAM and was disconnected from the equipment that failed. FAA documentation nevertheless says SMART was operating nationwide as an advisory platform. A detailed technical incident report remains pending.
- November 14, 5:15 a.m. ET — Recovery impact: Axios reported, citing FlightAware, that nearly 3,300 flights into, out of or within the United States were already canceled for Saturday and nearly 1,000 were delayed as airlines recovered. This is an operational consequence, not new evidence about the outage's initiating cause or any SMART connection.
- November 14, approximately 11 a.m. ET — Cause narrowed: The Associated Press reported that Duffy said a contractor's copper-to-fiber upgrade work at a remote facility led to a router overloading the network. He said officials were investigating contractor error or a faulty router. AP also reported his statement that more than 1,000 connection attempts every four to five seconds came from two locations. This is an attributed official account; the underlying logs and detailed incident report are not yet public. Yahoo News reported that Bedford said investigators would distinguish installation or configuration error from a defective router, that maintenance activity was recorded on paper, and that controllers shifted affected flights to working frequencies. A 2010 DOT inspector general review provides a close precedent: contractor configuration and procedural errors during a 2009 FTI fiber transition blocked about 75 percent of network traffic and delayed restoration for five hours.
- November 14 — Testing disclosure gap: The overloaded router identifies a failure point, but the public record reviewed by AINewsIQ does not show what generated the reported connection storm or provide a load-test plan, capacity model or results for SMART's nationwide rollout against legacy network capacity. This does not establish that testing was absent; it identifies evidence the FAA should publish. Mullin said new technology was overwhelming older systems; Duffy said new equipment remained online and SMART was disconnected. Duffy later raised possible L3Harris “incompetence” or a faulty router before a final cause was established; L3Harris said it was supporting the investigation; that statement did not accept responsibility or identify the failure. Campaign-finance records also show ASI’s PAC contributed $2,500 to the congressional campaign of Duffy’s son-in-law before the FAA announced ASI’s $875 million contract. None of this establishes procurement misconduct, improper influence or a SMART role, but it strengthens the case for independent technical and procurement review. The clarification also distinguishes remaining online from being incapable of contributing traffic to a shared dependency.
- November 14 — Recovery forecast: FAA Administrator Bryan Bedford said as many as 5,000 of Saturday's 41,000 scheduled flights could be canceled and expected the U.S. schedule to recover by the end of Sunday.
- November 14, 8:39 p.m. PT — Change-control review: The Associated Press reported that Professional Aviation Safety Specialists, the union representing FAA technicians, said the FAA normally consults its technicians before upgrades or repairs so they can assess risk, but that consultation did not occur for this work. This is an attributed union account, not an FAA finding. It strengthens the case for reviewing change approval and pre-work risk assessment; it does not establish that SMART caused or contributed to the outage.
- November 14, 8:39 p.m. PT — Operational status: DOT said more than 8,000 flight operations resumed after the ground stop lifted and that the airspace remained open, while AP reported the air traffic control system was operational and airline schedules were expected to recover by the end of Sunday. This updates recovery status, not the unresolved technical cause.
Editor's note
This is a living analysis. Material operational or investigative updates will be added with a timestamp rather than silently replacing the earlier record. AINewsIQ will distinguish official findings, attributed statements and unresolved questions. The central editorial rule is simple: state what has not been verified as clearly as what has.
Sources
- FAA: SMART expanding nationwide ↗
- FAA: Air Space Intelligence contract and SMART program ↗
- U.S. Department of Transportation: SMART operating model ↗ — official outage attribution and SMART/STARS/ERAM statement ↗
- Associated Press: nationwide ground stop and investigation ↗ — officials attribute the disruption to new technology overloading older systems ↗
- Axios: recovery disruptions continued into November 14 ↗
- Associated Press: contractor work and router investigation ↗ — technical context and connection-attempt count ↗ — DOT: November 14 contractor account ↗ — FEC: ASI PAC committee record ↗ — reporting on the contribution and family connection ↗ — Yahoo News: maintenance, paper records and controller response ↗ — DOT OIG: 2009 FTI outage review ↗
- Representative Don Beyer: October 9 pause request ↗
- House Transportation Democrats: GAO audit request ↗
- Nextgov/FCW: NATCA concerns and SMART oversight ↗
- FAA: FTI-2 program objectives and carried services ↗
- FAA: SWIM questions and answers ↗
- FAA: Newark telecommunications redundancy response ↗

Join the discussion
Have a different perspective or a question about the evidence? Discussion is free with a verified email account. Every comment is reviewed before publication.
No published comments yet.
New commenters: 2 submissions per 24 hours, at least 5 minutes apart, until 3 are approved. Discussion guidelines and moderation →