Officials ruled out AI and cyberattacks in a week of FAA, United and NSA outages, and Congress wants the evidence
By Albina Bieci, Staff Writer
Three unrelated disruptions, one week, and a public that cannot tell a bad deployment from an intrusion. Officials ruled out AI and cyberattacks; what they have not yet published is the root cause of any of it.

WASHINGTON
Federal officials dismissed rumors that artificial intelligence or a cyberattack caused recent disruptions across the FAA's airspace system, United Airlines and an internal National Security Agency communications network, according to reporting by CNN and Spectrum News, as Congressional scrutiny of interconnected U.S. networks intensifies.
The three disruptions were unrelated in origin and similar in effect: services that the public and the industry depend on stopped working, on the same week, for reasons that were not immediately explained. CNN's coverage makes the point that security officials now face a harder problem than intrusion detection. Cascading technical failures at the FAA, at major banks and at federal agencies look increasingly like coordinated attacks from the outside, and telling the two apart in the first hours requires information that is rarely available that early.
What has been confirmed
Federal officials stated that neither artificial intelligence nor a cyberattack caused the disruptions. The affected systems included the FAA's airspace system, United Airlines operations and an internal NSA communications network. Members of Congress have raised questions about the fragility of interconnected U.S. networks. CNN framed the broader issue as the growing difficulty of distinguishing glitches from deliberate attacks.
What has not been published is the technical root cause of any of the three events, the duration and scope of each, or the specific oversight actions Congress has proposed. Officials have also not described how the determination that no attack occurred was made, which is the detail a technical reader would most want.
Why the determination is harder than it sounds
Ruling out an attack requires evidence, not intuition. Investigators look for unauthorized changes to configuration, anomalous authentication, unexpected outbound traffic, the presence of tooling that has no business reason to be there, and a timeline that fits a deliberate act better than a failure. Each of those checks takes time, and each is harder on systems that were not designed to log the relevant events.
The alternative explanation is usually less dramatic. A configuration change pushed to a subset of servers, a certificate that expired, a dependency that returned an error under load, a routing table that converged incorrectly, a capacity limit reached during a traffic surge. Those failures produce the same user experience as an intrusion: a screen that does not load and a service desk that cannot say why.
The uncomfortable middle case is the one CNN's reporting describes. A failure can be caused by a mistake made in response to an attack, or an attacker can take advantage of an outage to move inside a network while attention is on restoration. Neither pattern is visible in the first hour, and both are common enough that investigators do not close the question when service returns.
Concentration is the underlying condition
The FAA, a major airline and a national security agency do not share infrastructure in any ordinary sense. They do share a set of conditions that make simultaneous-looking failures more likely than they used to be.
Commercial cloud platforms, identity providers, certificate authorities, content delivery networks and telecommunications carriers serve all three. A failure at any of those layers presents as an outage at every organization that depends on it, and the organizations themselves have limited visibility into the layer beneath them. Congress's interest in network fragility is an interest in that concentration, which is a procurement and architecture problem rather than a firewall problem.
The second condition is operational thinness. Organizations that reduced staff and consolidated tooling over the past decade have less capacity to diagnose a failure manually, and fewer people who remember how the system worked before it was automated. Recovery from a novel failure depends on that institutional knowledge more than on any dashboard.
The third condition is the pace of change. Continuous deployment improves resilience to small failures and increases the number of changes that can produce a large one. An organization that ships hundreds of changes a week has hundreds of opportunities to introduce the kind of defect that surfaces only under specific conditions.
What Congress is likely to ask for
Oversight of an outage tends to follow a predictable sequence. Members request a briefing, then a written timeline, then the root-cause analysis, then a description of what changed as a result. The productive version of that sequence produces three artifacts: a dependency map that shows which external services each critical function requires, a record of the last test of the manual fallback for each of those dependencies, and a documented decision about which failures the organization is willing to accept.
Those artifacts are unglamorous and they are what distinguishes an organization that recovers in hours from one that recovers in days. Most agencies and most companies can produce the first one in outline. Fewer can produce evidence that the fallback was tested.
The physical side of a cyber or operational outage
Buildings do not stop needing protection when the network does. Modern facilities route fire alarm monitoring, access control, elevator communications, environmental monitoring and emergency notification through systems that depend on connectivity and on staff who can operate them.
When those systems go down, the building's protection is impaired, and the consequence is the same one the fire code assigns to a failed sprinkler system: someone has to walk the property, on a schedule, with a written record, until the system is restored and tested. That is a fire watch, and it is ordered for the same reason whether the cause was a broken pipe or a failed network link.
Data centers, hospital campuses, high-rise office towers and industrial sites share this exposure, and the organizations that operate them rarely rehearse the handover from automated monitoring to physical patrol. The transition takes hours to arrange on a normal day, and an outage is not a normal day.
What operators should do before the next one
Write the dependency list at the level of function, not vendor. Which functions stop when the identity provider is unavailable, which stop when the carrier fails, which stop when the cloud region is degraded. The list should be short enough to read in a meeting and specific enough to act on.
Test one fallback per quarter, and record the result. A manual alarm response procedure that has never been executed is a document, not a capability. The test should include the notification path to the monitoring company and to the local fire department, because both have their own procedures and neither will be waiting for a call.
Decide in advance who speaks publicly and what they will say. The vacuum created by a slow explanation is filled by inference, and inference in a week with a ransomware listing in the news is unkind.
And separate the two clocks. Restoration is an engineering problem with an unknown duration. Communication is a scheduled obligation that can be met whether or not the engineering is finished.
Related coverage
USPA reported this week on Visa's confirmed third-party incident and the ransomware listing that arrived beside it, where the same attribution problem played out in the private sector.
On the exposure that precedes an intrusion, the FBI's flash advisory on automated reconnaissance of corporate facilities describes how much of an organization's technical footprint can be mapped before anyone notices.
For the program view, USPA's corporate risk intelligence practice covers continuity planning, dependency mapping and the physical coverage that continues when building systems do not.
Two monographs in USPA's research library bear on this week. The National Capital Region risk analysis covers the federal district's infrastructure and its interdependencies, and the subsea cable protection analysis covers the physical layer carrying the connectivity every one of these systems depends on. On the operator side, the grid security and storm surge logistics case study is the closest published example of scaling physical coverage during a regional outage.
The question that remains open
Federal officials have said what did not cause these outages. The public record does not yet say what did, and until it does, the useful posture for any organization that depends on the same shared infrastructure is to assume the failure mode is available to them as well.
Why the public conversation keeps landing on the same question
Every one of these events produces the same three-day arc. Service fails. Users report it broadly and simultaneously. Someone with a large following speculates about an attack. Officials respond that there is no indication of one. The speculation outlives the correction, because the correction is a sentence and the speculation was a narrative.
That arc is not a failure of journalism so much as a structural feature of the information environment. Attribution requires access to evidence that only the operator holds, and the operator is legally and operationally constrained in what it can release while an investigation is open. The gap is filled by inference, and inference during a week with a ransomware listing in the news runs toward intrusion.
The remedy is not better speculation. It is operators publishing the operational cause quickly when it is mundane, in enough technical detail to be checkable, and being specific about what remains unknown.
Readers who manage facilities can use the week as a prompt for one specific check: whether the building's protection systems have a documented degraded mode that does not depend on the corporate network. If the answer is a blank page, the work is to write it and test it, because the failure mode under discussion is available to a mid-size office building as readily as to a federal agency.
USPA Nationwide Security provides fire watch and physical security coverage for facilities whose monitoring or protection systems are offline, including data centers, hospital campuses and high-rise properties. The dispatch desk is answered 24 hours a day at (800) 214-1448.
Services in Washington, DC
Sources
- CNN — reporting on the outages and the difficulty of distinguishing glitches from attacks
- Spectrum News — reporting on Congressional scrutiny of interconnected networks
- Federal Aviation Administration — agency information
- Cybersecurity and Infrastructure Security Agency — critical infrastructure guidance
- National Institute of Standards and Technology — contingency planning publications
About the reporter
Staff Writer
Staff writer for the USPA Nationwide Security newsroom. Receives newsroom tips, corrections and editorial inquiries.
Author biography and credentialsMore from The Watchline

Industry
USPA expands its cybersecurity practice to website security upgrades and secure rebuilds
The practice now scopes hardening work on existing sites alongside full secure rebuilds, hosting and email migrations, and continuing management for businesses whose websites have outgrown the platform they were built on.
· Albina Bieci

Industry
Coast Guard cutter Tahoma offloads 3,622 pounds of cocaine at Port Everglades after a Caribbean interdiction
Four detainees were handed to federal agents and roughly $27 million in cocaine came ashore, the visible end of a detection problem that starts far offshore.
· Albina Bieci

Industry
USPA Nationwide Security appoints Michael Evans Jr. Associate Manager of Maritime Operations
The New York-based appointment follows progressive vessel assignments and maritime studies at Maritime Professional Training in Fort Lauderdale.
· Albina Bieci




