A rough few days for technology reveal our digital vulnerabilities

Tech has gone through a tough few days. Between Thursday, November 12, and the following Tuesday, we saw the resignation of Nvidia CEO Jensen Huang, along with at least 11 reported significant tech outages. They included a widespread disruption of Microsoft services, including Outlook email; a nationwide FAA grounding of departing U.S. flights; disruptions to bank money transfers; a roughly two-hour disruption of United Airlines departures; outages affecting Amazon Web Services and several Apple services and an outage of the National Security Agency’s highly classified internal communications network, known as NSA Net.

I included Huang’s resignation because, despite the obvious differences, it has two things in common with some of the technical outages. First, a single point of failure, whether a person, a piece of equipment or a software glitch, can have enormous consequences, both in economic losses and damage to public confidence in the companies and agencies we depend on. Billions in stock value were lost, at least temporally, not just by those who own Nvidia stock directly, but also by millions of investors who own it indirectly through mutual funds and exchange-traded funds (ETFs). Second, Huang has been one of the world’s leading proponents of AI, having built Nvidia into the dominant supplier of chips that power today’s AI systems. And regardless of whether AI had anything directly to do with these crises, it became a central part of the public discussion surrounding them.

FAA outage disrupted travel nationwide
The FAA failure had perhaps the greatest impact on Americans, grounding departures nationwide for nearly five hours. The FAA reported more than 12,000 flight cancellations on November 13. Disruptions continued over the following days, affecting travelers, their families, employers, businesses and events that depended on them getting where they were going. Air freight was also disrupted.

I’m not going to add my voice to the blame game over who was responsible. But according to the FAA, a newly installed router at a remote facility in Arkansas generated excessive network traffic that overwhelmed older equipment, affecting 17 of the nation’s 22 regional air traffic control centers. The problem raises questions about whether the system adequately followed a fundamental principle established by the original designers of ARPANET, the predecessor of today’s internet. Back in the late 1960s, they built redundancy into the network so that data could travel along multiple paths, routing around failed connections or equipment, rather than allowing a problem at a single facility to create conditions that warranted shutting down a critical nationwide system. In this case, a problem involving one router exposed vulnerabilities in an aging nationwide network.

When digital systems fail
Even though it was short-lived and had limited impact, the failure at Chase, which also affected other banks, is a reminder of how dependent we are on digital data. I’m old enough to remember having a “bank book” that showed exactly how much money I had on deposit. Today, the only up-to-date record of how much I have is digital, and even though I’m assured there are backup systems, I’m not immune to worrying about the remote possibility that those records could somehow disappear. I don’t know if it’s necessary, but I occasionally download my financial statements and print them on paper, just in case.

Even the National Security Agency wasn’t immune to technical problems. Its classified internal communications network experienced disruptions during routine maintenance, affecting internal systems. The NSA said its mission-critical operations continued, thanks in part to redundant systems.

AI and the risks of updating systems
Government officials have said that there is no evidence that the major outages were caused by cyberattacks or that AI was directly responsible. But AI may have played an indirect role. Security researchers are increasingly using AI to identify vulnerabilities, leading to software patches that can have unintended consequences. At the NSA, software maintenance reportedly contributed to its network problems, although it’s unclear whether any specific vulnerabilities being addressed were discovered with AI.

In a statement, Sen. Mark Warner, the top Democrat on the Senate Intelligence Committee, said, “AI is making it possible to identify vulnerabilities across enormous numbers of systems at unprecedented speed, creating enormous pressure on companies and government agencies to patch those vulnerabilities immediately, often all at once, putting tremendous strain on systems that weren’t built to handle that pace of change.”

It’s a known fact that updates, even when motivated by increased security, can sometimes introduce new problems into systems. Still, I recommend that you keep your devices and software up to date. The risks of not updating are generally far greater that the possibility of a new bug.

AI is helping both sides
It’s a good thing that AI can help identify and fix security vulnerabilities. But cybercriminals and nation-state hackers are also using AI to plan and carry out attacks, creating a cat-and-mouse game between attackers and security professionals.

Recent cyberattacks on supermarket refrigeration systems in Europe show that hackers can disrupt more than computers. And the risks are even more troubling now that the bad guys have increasingly powerful AI tools.

Google’s latest  AI Threat Tracker report, warns that cybercriminals are using autonomous AI systems to search for software vulnerabilities, examine updates to figure out which flaws were fixed, and develop ways to exploit those flaws on computers that haven’t yet been updated, without a human directing every step. “Imagine a hacker that doesn’t eat, sleep, or get tired,” it says, “and never stops searching for a way in.” And AI is not only being used by private sector criminals but also nation states who can put considerable resources into attacking other countries.

The continuing need for human expertise
Even though AI can help companies and agencies identify and address vulnerabilities, we are far from the point where humans can be taken out of the loop. We still need programmers, systems engineers, cybersecurity specialists and IT professionals to prevent and fix problems.

Unfortunately, the federal government and many private companies have cut back on personnel, including some involved in maintaining critical systems. In 2025, the FAA dismissed fewer than 400 employees, including some maintenance workers, while reports indicate the NSA reduced its workforce by approximately 2,000. Microsoft announced 4,800 job cuts in July 2026, and Amazon eliminated another 16,000 positions in January, including jobs at AWS.

I don’t know if these staffing reductions caused the recent outages, but they raise questions about whether companies and government agencies have enough experienced people to maintain, protect and repair the increasingly complex systems we depend on.

I’m glad that lawmakers are now calling for investigations into the resilience of both government and private-sector networks, but we need more than just congressional hearings.  It’s time for government and business leaders to devote the resources needed to protect our nation’s digital infrastructure, which is increasingly interconnected with critical systems, including transportation, energy and health care.