Flights stopped in America and Microsoft shut down: Did artificial intelligence play a role?

The decision of the US Federal Aviation Administration (FAA) to suspend flights nationwide yesterday, due to a communications problem that affected several air traffic control centers, coincided with a malfunction that affected Microsoft 365 services and began the day before, raising questions about the possibility of a connection between the two incidents. But the available evidence does not prove that there is a common cause behind them, while the underlying cause of the communications problem remains uncertain.

This leaves a real gap in the publicly available record. The two incidents coincided, and artificial intelligence products were affected, but neither proves a connection. The possibility of a link remains if investigators find evidence of a common dependence or chain of faults.

What happened? And when?

The Federal Aviation Administration said in a statement issued on November 13 that it had issued a decision to suspend flights nationwide, after communications problems affected several air traffic control centers on the air routes. She added that technicians are investigating the problem, and described the available information as preliminary.

Stopping flights means keeping planes scheduled to take off at the relevant airports on the ground. Travelers should check with airlines and airports for information regarding their flights; Issuing a nationwide warning does not specify the departure time of a specific flight.

The Associated Press reported that flights departing from Europe to the United States were suspended, at the request of the Federal Aviation Administration, from November 13, at 22:00 UTC (5:00 p.m. EST), until November 14, 05:00 UTC (midnight, EST).

The Associated Press quoted Homeland Security Minister Markwayne Mullen's explanation, who attributed the problem to an increased load on the system, noting that there were no indications of an attack. Transport Minister Sean Duffy said the SMART system was not linked to the malfunction. But the ultimate cause of the problem has not yet been confirmed.

FAA times indicate when notices will be issued. The chronology is important because it shows that the two events did not begin at the same time. Simultaneous failures may have a common cause, but timing alone does not reveal a connection between them.

Questions remain

Microsoft updates published by the Penn platform attribute the disruption to changes in the network, and deny that it is a cyber attack or that it is related to independent artificial intelligence agents.

The revised explanation does not reveal who wrote the code. Linking the bug to code written by AI requires access to development logs.

Microsoft also told the Associated Press that it has no indications of a link between the suspension of flights announced by the Federal Aviation Administration and the disruption that struck its services.

Impact of artificial intelligence products

The clearest documented link to AI is the vulnerability of these products to malfunctions. Microsoft has included the Copilot Chat service and access to work data among the affected jobs. This does not mean that all Copilot functions, all Microsoft services, or the Azure cloud will be completely disabled.

Anthropic said that the Microsoft 365 problem caused malfunctions in the service connector for its assistant, Claude, as well as in Claude for Outlook. Its subsequent report identifies the main impact period from November 12, at 16:03 UTC (11:03 a.m. EDT), until November 13, 16:40 UTC (11:40 a.m. EDT).

Anthropic later announced that the problem had been resolved, while warning that occasional errors or delays may persist while Microsoft services recover.

OpenAI reported problems establishing new connections to the Microsoft 365 connector and the connector tools, before moving the incident to the monitoring phase. The company's official incident data records that the issue was resolved on November 14, at 00:18 UTC (7:18 PM EST on November 13). OpenAI said that the affected services had returned to operating normally.

The connector allows an AI product to request information or perform a task authorized through another service. If that service or connection goes down, an email or calendar task may not be possible, even if the AI ​​interface itself remains responsive. These reports relate to specific connections, and do not prove a general malfunction in Cloud, ChatGPT, or artificial intelligence models.

Why is there controversy over the SMART system?

The FAA's SMART system is based on a single site near Washington, where a team of experts analyzes flight schedules, weather forecasts, and hundreds of other sources of data. This office then sends recommendations to the FAA Command Center, which then transmits them to air traffic controllers across the country for consideration. The company “ASI”, which designed this system, stated that “the current decision to stop flights is not based on the SMART system, nor is it a result of it, nor is it related to any ASI product or its integration process with the Federal Aviation Administration (FAA) and the air traffic control system (ATC),” according to the Associated Press.

The Federal Aviation Administration recently expanded the use of an AI-powered planning system called SMART. The administration unveiled the system on September 21, and announced on November 10 that it would expand its scope to include the airspace of all 22 centers. Live coverage by the Associated Press also reported that Mullen said that there are so far no indications of a connection between the system and the malfunction.

Air traffic controllers remain responsible for ensuring the separation of aircraft. This distinction is important; Planning software can influence decisions, which makes testing it necessary, but the declared role of the system does not mean that it is an independent alternative to monitors.

The Federal Aviation Administration awarded Air Space Intelligence a contract to provide air traffic flow management services and data (FMDS), and described the SMART system as a development of these services. The administration says that FMDS aims to replace the old air traffic flow management system and help achieve a balance between demand for flights and available capacity. These documents describe efforts to modernize the systems, but they do not identify “SMART” as the component responsible for the communications problem that occurred yesterday (Friday).

It also does not prove that SMART uses Microsoft 365, or that Microsoft is the provider of its cloud services or identity system. There may be a common dependence between a system based on cloud computing and a malfunction in cloud services, but this dependence must be identified and verified, and it cannot be inferred simply because the word “cloud” appears.

In a statement issued on November 13, Air Space Intelligence denied that the SMART system, its other products, or its integration with the Federal Aviation Administration and air traffic control centers caused the grounding of flights. This remains the story of the company providing the system, and is not the result of an independent investigation to determine the root cause of the malfunction.

What evidence can prove a link between the two incidents?

Proving a credible connection requires more than just that the two incidents occurred on the same day. Investigators must identify the component that failed at the FAA, and prove its dependence on a specific service, network, identity system, or telecommunications company that was party to the Microsoft failure. Logs of interconnected systems, logs of changes, and documentation of the path that led from this dependence to the breakdown of communications will be stronger evidence than mere coincidence of events.

The existence of a common dependence may lead to the impact of the failure being transmitted from one system to another, without this meaning that there is a single root cause for the two incidents. But the existence of such a chain has not yet been verified. The final technical results of the Federal Aviation Administration and the final review of the accident by Microsoft are still not available, according to American media.

Allegations of an AI-directed attack require evidence of their own.

Planes stopped taking off from American airports, and air travel was disrupted across the country, after a sudden malfunction struck the communications systems on which air traffic control depends.

While the authorities were able to contain the malfunction and reopen the airspace, the accident raised questions about the fragility of the technical structure of American aviation, especially with parts of its systems relying on equipment dating back decades.

US Secretary of Transportation Sean Duffy warned of the possibility of similar malfunctions recurring unless these systems are updated, at a time when the repercussions of the crisis extend beyond airports to affect travelers and other transport companies.

The Federal Aviation Administration explained that the decision to stop takeoff came after communications problems that affected a number of air traffic control centers, while Duffy said that the disturbance began around 3:36 p.m. local time, and led to instability in the communications and radar systems.

The authorities lifted the decision to stop flights at 8:15 pm, after restarting the affected centers and facilities, with delays and cancellations continuing as a result of the accumulation of disturbances.

What caused the flights to stop?

At a White House press conference, Duffy attributed the flaw to the outdated communications systems used by the Federal Aviation Administration, explaining that the network relies on outdated analog technology known as “TDM,” along with routing switches, routers, and copper lines dating back decades.

The minister said that these devices were subjected to great pressure due to the data, and began sending inconsistent information, which prompted the authorities to stop the take-off movement as a precaution in order to preserve aviation safety.

He added that the technical teams were able to relieve the pressure on the routers and restart the network, noting that the newer equipment that was installed did not suffer from a malfunction.

According to the Associated Press, the authorities have not yet provided a final explanation for all the circumstances that led to the network disruption, but they have suggested that the problem is related to the interference of old systems with the process of modernizing the infrastructure.

Other malfunctions may occur

The most prominent warning in Duffy's statements related to the possibility of a repeat of the incident; When asked whether Americans might face similar disruptions again, he said that the old system was still vulnerable to such problems, adding: “You will see more of these disruptions” unless the government and Congress provide the necessary funding to replace the network.

He explained that he could not anticipate all the major malfunctions that might affect the analog communications network, nor confirm that the same scenario would be repeated, but he pointed out that the United States had witnessed previous problems in New York and Newark, and that the authorities were still working to determine precisely what happened and how to prevent its recurrence.

Duffy said that technical investigations are continuing, and that the specialized teams are working to find out the causes of the defect and develop procedures to prevent similar problems from occurring in the future.

Upgrading an old network requires billions of dollars

The accident has brought to light the project to modernize air traffic control systems in the United States, parts of which rely on old equipment, despite the fact that the authorities have begun replacing some fiber-optic lines with copper communications lines, and modernizing wireless communication devices in a number of airports and control centers, according to the Associated Press.

Duffy said that the available funding is not enough to complete the modernization process, calling on Congress to provide additional resources. He estimated that completing the project would require about $50 billion, while he indicated that the administration was seeking additional funding amounting to $30 billion.

He stressed that replacing modern systems with old ones, which include hardware, software and communications networks, would reduce the recurring malfunctions in the current infrastructure.

Disruption in Amtrak services with increased demand for trains

The repercussions of the suspension of flights extended to the railway transportation sector, after Amtrak, the national train operator in the United States, announced an unusual increase in customer communications, as a result of travelers seeking alternatives to the affected flights.

The company said that its website and application encountered error messages, and that reaching customer service representatives had become more difficult due to pressure on communication lines, calling on travelers to try again later. However, it confirmed that the train services themselves were operating normally, and did not announce any interruption or disruption in operation.

Was the flaw a cyber attack or related to artificial intelligence?

While the suspension of flights raised questions about the possibility that aviation systems were exposed to a cyber attack, US Secretary of Homeland Security Markwayne Mullen said that there were no indications that the incident resulted from an attack of this type, stressing that the investigations did not reveal evidence of sabotage activity, according to the Guardian newspaper.

The authorities also denied that the new “SMART” air traffic management system, which the Federal Aviation Administration announced was activated in air traffic control centers, was a direct cause of the malfunction. Duffy explained that the new systems were not affected, and that the defect affected the old communications equipment.

However, the authorities have not announced the end of the investigation into all aspects of the incident, and have not yet provided a final explanation as to why the old network was loading data beyond its processing capacity.

Previous failures raise concerns about system integrity

The disruption of US aviation systems is not unprecedented; In January 2023, the United States witnessed a widespread cessation of domestic flights, after a defect in the pilot alert system known as “NOTAM” caused the suspension of takeoffs for hours. Airports in New York and Newark also faced problems in communications and air traffic control systems that led to widespread delays.

But the latest accident took on a broader scope, after its impact extended to air traffic control centers nationwide, and led to the suspension of takeoffs from American airports for more than 4 hours.

The Associated Press reported that more than 3,000 planes were in the air during the crisis and were able to land safely, while hundreds of flights were diverted, and the travel schedules of large numbers of travelers were damaged.

Despite the reopening of airspace, Duffy warned that restoring air traffic to usual levels will take time, due to airlines having to reallocate aircraft and crews, and deal with accumulated delays and cancellations.