Our brand is chaos

Due to lawyers’ exces­sive depen­dence on all things Microsoft, the rule of law is taking hits from this week’s Outlook outage. (I was not affected—I use Fast­mail.) But weird cyber­se­cu­rity events are becoming more common. For instance—last week, Vander­bilt Univer­sity was hacked. And last month, commer­cial refrig­er­a­tors in France were too.

AI is indis­putably revealing computer-secu­rity flaws that humans have long over­looked—for instance, last month AI discov­ered a crit­ical vulner­a­bility in PHP, a program­ming language used on 70% of websites, that had been lurking for 10+ years.

Still, on a net basis, is AI improving computer secu­rity? That’s going to be a thornier ques­tion, because AI’s ability to reveal secu­rity flaws is coun­ter­acted by two other AI effects:

  1. Hackers—some state-affil­i­ated—are using AI tools to find and exploit weak­nesses in systems before they can be detected or fixed, at lower cost than ever.

  2. The rise of AI-powered coding tools within soft­ware orga­ni­za­tions is quietly intro­ducing weird new vulner­a­bil­i­ties that elude human detec­tion.

When we think of AI-powered mischief, we may imagine hackers cine­mat­i­cally targeting banks and secret govern­ment servers. But these systems are oper­ated by enti­ties with big budgets who can afford high-end secu­rity and rebuff casual attacks. The greater risk is that AI-powered hacking will dispro­por­tion­ately accel­erate against softer targets with smaller secu­rity budgets—espe­cially schools, hospi­tals, and crit­ical infra­struc­ture like water and power. It’s already happening.

Was the Microsoft outage AI-related? No confir­ma­tion yet. Though just last month, Microsoft described an AI-driven cyber­at­tack with similar char­ac­ter­is­tics. In general, I expect that AI-powered hacks and outages will become a more regular feature of Amer­ican life, and we will have to navi­gate around the failure of systems so far notable for their mundane reli­a­bility.

As certain AI models are used for attacks and certain others for defense, we can foresee a future where the severest cyber­se­cu­rity vulner­a­bil­i­ties—and their fixes—will both be AI gener­ated, and too complex for humans to compre­hend. Once we pass that threshold, we will be observing the first largely autonomous AI-vs.-AI conflict, where humans are inca­pable of under­standing the nature of the battle, yet will be enduring severe conse­quences.

update, 0 days later (13 November 2026)

Liter­ally minutes after I finished this piece, I saw that the FAA has today imposed a nation­wide ground stop due to “commu­ni­ca­tions prob­lems” with certain ATC systems. AI is not offi­cially the culprit, but Gary Marcus—stuck at EWR—notes that the FAA recently rolled out nation­wide its new AI air-traffic routing system called—smh—SMART.

update, 3 days later (16 November 2026)

The Microsoft outage sure sounds like the result of an AI-gener­ated soft­ware patch being pushed onto live servers without being fully tested or under­stood. This is the same company that recently offered buyouts to 27% of its work­force because of the expected effi­cien­cies of AI. The path from here leads only one direc­tion. Microsoft soft­ware quality has been a punch­line for decades. But in the desktop-soft­ware era, one could always refuse to install an update that was reported to be buggy. These days, with desktop soft­ware becoming increas­ingly a wrapper around cloud services, this avoid­ance tech­nique becomes impos­sible: if the cloud breaks, so does every­thing that depends on it. It is gobs­macking that a public company that’s already squan­dered its once consid­er­able lead in AI is now seeking to throw its existing, stable busi­nesses onto that fire, along with its work­force. But the effect on Microsoft soft­ware quality is easy to predict: more outages, lasting longer. (GitHub users have already witnessed this down­ward spiral for months.) Further­more, this week’s outage revealed that many Microsoft customers have no alter­na­tive email systems.

update, 3 days later (16 November 2026)

A JPMorgan Chase soft­ware bug is impairing banking trans­ac­tions. Sepa­rately, United Airlines expe­ri­ences its own system outage.

update, 4 days later (17 November 2026)

Apple is expe­ri­encing an outage affecting many cloud services. Sepa­rately, an NSA commu­ni­ca­tions network is down. A US Senator says “If this is all a coin­ci­dence … it’s like the greatest coin­ci­dence in the history of the world.”

update, 4 days later (17 November 2026)

I enjoyed The Deco­her­ence Trap, a piece that describes a specific kind of rot that AI-gener­ated code will be suscep­tible to.