Investigators weigh coincidence, shared technology and external intrusion

A string of major technology failures over one week disrupted systems and services that millions of Americans rely on every day, raising questions about their reliability and security. The latest incident involved the National Security Agency, which said its own internal systems were affected by an outage that day; the outage happened as part of routine maintenance and did not affect security operations. Several news reports said the outage involved a classified internal network.

The sequence included a Microsoft outage last Thursday that disrupted services around the world and a nationwide FAA ground stop on Friday that brought air travel to a halt. United Airlines later experienced a separate outage, while customers of Chase, Bank of America and Wells Fargo reported problems accessing their accounts. The incidents were said to be unrelated, but their timing has fueled concern that a common factor may be involved.

Eric O’Neill, a former FBI counterintelligence operative who is now a cybersecurity executive specializing in cyber threats and national security, said no evidence currently connects the failures beyond a massive coincidence. He said the organizations and government agencies involved have not been forthcoming, limiting what can be learned about the outages. O’Neill added that other companies beyond those named had also gone down. He said the coincidence was too strong to dismiss a possible external factor and that investigators would look for one diligently.

Homeland Security Secretary Markwayne Mullin, interviewed the previous night, dismissed suggestions that a coordinated cyberattack was responsible and instead blamed artificial intelligence. Mullin said he had consulted the technology industry, including an unnamed adviser who asked not to be named and was not Elon Musk. The adviser said he believed AI was responsible because programs were not designed to run as quickly as AI. He said code able to write and fix itself creates a ripple effect that can overwhelm systems not built for that speed. PBS NewsHour characterized the explanation as far from conclusive.

O’Neill said Microsoft called its disruption a technical glitch. He described the air-traffic event in greater detail, saying one of 21 nodes in the air traffic control communications system began communicating erratically and that this led to the FAA grounding all flights. O’Neill guessed that the FAA acted under an abundance of caution because officials believed they faced a cyberattack, but he presented that motive as his own speculation. He described the NSA incident as a problem involving one of its classified systems.

The former FBI operative said investigators could not be confident at all about ruling out a sophisticated intrusion. He described three lines of inquiry: a coincidence of internal failures among cloud-using companies that are broadly used across the industry; an external attacker or force that came in and caused a flaw or disruption; and a common application, networking tool or piece of technology used across similar industries that broke down, went bad or was attacked by an external threat actor. O’Neill said the affected cloud-using companies are part of critical infrastructure. He said investigators might find the shared-technology explanation more likely, but all three lines of inquiry could produce the same disruption. He expected investigators to have substantially more data by the following week.

Asked whether Mullin’s artificial-intelligence explanation held up technically, O’Neill said it did. He then argued that the AI capable of overwhelming systems was coming from external attackers rather than from inside the affected companies. He identified foreign intelligence services and cybercrime syndicates as users of AI agents that could keep probing systems without pause, forcing defenders to operate continuously as well. O’Neill said the cybercrime syndicates were making trillions of money from these attacks.

O’Neill said cyberattacks in this quarter of 2026 were 53 percent more numerous than at the same point last year. “There will always be vulnerability. But the threat is worse than it’s ever been.” He said the increase required greater resilience and more cybersecurity investment.

He said the cybersecurity of the future would differ from what had been used for decades, using AI against the best AI tools available to criminals and spies. He said that approach was already being deployed and built and was what would save systems. O’Neill is the author of Spies, Lies, and Cybercrime: Cybersecurity Tactics to Outsmart Hackers and Disarm Scammers.