Non-NSA outages were not cyber-attacks and did not appear AI-related; NSA case unknown, Schneier says
In an opinion column published Nov. 20, 2026, in The Guardian’s Comment is free section, security technologist Bruce Schneier rejected the idea that a week of U.S. technology failures shared a cause. The outages he listed ran to Microsoft, the FAA, a handful of banks, United Airlines, an unnamed classified NSA network and Apple. For the incidents apart from the one at the NSA network, Schneier wrote that they were unrelated, were not the result of any cyber-attack, coordinated or not, and did not seem to be the result of artificial intelligence. He said nothing could be known about the NSA case and that its details might never be known. His explanation for the infrastructure’s fragility was economic rather than a shared attack: the systems are built and maintained in a way that turns reliability into a cost instead of a source of near-term profit, and changing that requires regulation with enforcement and meaningful fines, and stringent antitrust enforcement.
His thesis is explicit: “The technologies we rely on are built and maintained within our economic system – and that’s where the core problems lie.” He wrote that the technological reasons for the condition of the infrastructure are all downstream from the economic ones, and framed the problem as a product of the incentives baked into shareholder capitalism. He described the outages he could assess as a visible instance of a wider condition rather than the work of a common attacker.
Schneier said people should be forgiven for suspecting a coordinated cyber-attack, a common cause or at least a conspiracy, but argued that the cases he could assess were unrelated and not the result of any cyber-attack, coordinated or not. On the NSA case, he was more cautious: “We can’t say anything about the NSA; we’ll probably never know the details of that one.” He described the incidents he could assess as “really just a series of unfortunate coincidences,” and said those incidents should not have surprised anyone because the systems behind them are “both vulnerable and fragile.”
He described that condition in plain terms: “Our computerized infrastructure is both vulnerable and fragile.” Much of it, he said, is old and has been incrementally patched and upgraded over the decades; a surprising amount of it is maintained by hobbyists, without any recompense; there often aren’t good backup systems; and more and more of it is centralized, meaning that individual failures have cascading consequences across many different aspects of our lives and society. “A lot of our tech infrastructure just barely works,” he wrote, and when it fails it can fail badly — “the very opposite of resilience.”
The technical weaknesses, he argued, have economic causes. Security, safety, resilience, redundancy and recoverability are all essential parts of reliable technology, Schneier wrote, but they are all costs, and more importantly they are costs that don’t generate near-term profits. Modern shareholder capitalism, and private equity in particular, prioritizes near-term profits over everything else, he wrote, squeezing all apparent inefficiencies — and, additionally, all other positive intangibles — out of a company in pursuit of those profits.
He also counted competition among the things that get squeezed out, though in a narrower sense than the rest. Market competition, he wrote, is essential for the forces that nudge companies to provide better services, and at the same time it is another perceived inefficiency. “Monopolization removes it in pursuit of near-term profits.” The quarterly-budget view of reliability spending follows the same logic, he said: from the perspective of a quarterly budget, spending on reliability measures is an inefficiency, an expense that doesn’t generate revenue, and it feels wasteful until something bad happens.
When the costs of failure can be pushed elsewhere, they are pushed onto society, further increasing profits, he wrote, and the bill arrives in public: “we get stranded in airports when planes don’t fly, or have to deal with the fallout when we can’t withdraw money from our bank accounts.” The effects of those economic pressures, he wrote, are brittle systems with single points of failure and cascading failure modes, more susceptible to cyber-attack — both human and AI-fueled — and more susceptible to random mistakes. The more unregulated capitalism is allowed to privatize gains while socializing risks, he argued, the more such failures should be expected.
His remedies address the incentives rather than the equipment. Changing what makes it profitable to operate these systems just at the edge of functionality, and to push the costs of failure on to everyone, requires regulation with enforcement and meaningful fines, he wrote, and stringent antitrust enforcement. Upgrading computerized systems, increasing their reliability, removing single points of failure and making them more resilient are all technical problems, he wrote, but they will not happen until the political and economic problems are fixed.
He was less confident about implementation. While the prescription is easy to state, he wrote, it is much harder to implement — especially in the US, where “monied interests have such a tight hold on the legislative agenda.” Until that changes, he wrote, we should expect lots more of these sorts of failures in the future. Schneier is a security technologist who teaches at the Harvard Kennedy School at Harvard University and the University of Toronto’s Munk School.