NPR’s Sacha Pfeiffer speaks with cybersecurity expert Dmitri Alperovitch about the coincidence of several major technical flaws in IT infrastructure and the best way to understand them.
Transcription
PRESIDENT DONALD TRUMP: It’s life. It’s life. When you’re big and powerful and have all the money, things like that happen. Alright.
PFEIFFER: El presidente Trump restó importancia a todos los apagones de hoy. Aunque, hablando de ser grande y poderosa, NPR ha confirmado que la red clasificada que la Agencia de Seguridad Nacional utiliza para sus propias comunicaciones internas ha estado teniendo problemas hoy. Tenemos tantas preguntas sobre todo esto, por eso hemos llamado a Dmitri Alperovitch. Es cofundador de la empresa de tecnología de ciberseguridad CrowdStrike y fundador del Instituto Alperovitch de Estudios de Ciberseguridad de la Universidad Johns Hopkins. Thank you for coming on the program.
ALPEROVITCH: No. Y entiendo que la gente desconfíe mucho. Obviously, all these events happening almost every day is very strange. But these things happen. Y, ya sabes, una de las leyes de Murphy, por supuesto, es que nunca se debe atribuir a la malicia nada que se explique adecuadamente por la incompetencia. Creo que muchas de estas cosas suceden porque la gente está en medio de muchas mejoras en este momento.
There is, of course, the AI threat. These AI models are finding many vulnerabilities. And as a result, many companies and organizations around the world are trying to fix these vulnerabilities as quickly as possible. So we’re going through this patch apocalypse right now. And sometimes when you upgrade systems, we all know that things can go wrong, and I think that’s partly what’s happening here.
PFEIFFER: También estamos sufriendo esta colisión de nuestra extrema dependencia de la tecnología, sistemas sobrecargados ya menudo obsoletos que, como usted dijo, necesitan actualizarse periódicamente. We know that cyberattacks are common. We know that rogue AI is a fear. ¿Estamos viendo cómo se une la fragilidad del sistema?
ALPEROVITCH: Well, these are all different systems. So there is no doubt that we depend much more on technology. So when things go wrong now, it causes us much more inconvenience. You know, in the old days, maybe it was easier to go into your bank and physically withdraw money, and now we rely a lot more on online banking and that kind of thing. So the impact for us is much greater, but these things have always happened. But I do think there’s something to the idea that, right now, it’s happening more because everyone is trying to update their systems before these AI models can hack them.
ALPEROVITCH: That’s right. More vulnerabilities have been discovered this year than in the last 30 years combined. That’s because AI analyzes these software packages and identifies new vulnerabilities. So companies and government agencies are, right now, completely inundated with reports that they need to update their systems because another critical vulnerability has just been discovered.
PFEIFFER: Es como un juego del gato y el ratón en términos de tecnología, actualizaciones y malos actores que buscan vulnerabilidades. So does that suggest that this will simply be the ongoing game of cat and mouse that lasts forever?
ALPEROVITCH: Well, that’s what life in cybersecurity has been like for the last 40 years, and I don’t think it’s going away. As long as there are people with malicious intentions, whether nation states, criminals or hacktivists, we are going to have problems. And, of course, the greatest vulnerability is still the user, not their computer system, because people fall for social engineering attacks, for phishing scams, and I hope it continues to be that way.
PFEIFFER: Y las personas tienen contraseñas débiles y es posible que no utilicen la autenticación de dos factores. That? Para dar algunos consejos prácticos a la gente, ¿somos todos víctimas indefensas de esto, o qué cosas básicas deberían hacer las personas para tratar de protegerse?
ALPEROVITCH: Well, in terms of your own systems, yes, update it automatically. Make sure you have the full patch every time a vulnerability arises. Make sure, like you said, you have two-factor authentication and strong passwords. But when it comes to the failure of this infrastructure, we are all observers here and unfortunately sometimes victims because if your bank fails, there is not much we can do.
ALPEROVITCH: Well, there usually is, and it was a surprise that there was some kind of weakness in the system that could ruin everything. Usually, there’s redundancies. Now, the problem is that many of these systems have become so complex, with so many layers over the years, that it is very, very difficult to understand where those points of failure might be. And hopefully this is an area where AI could help us identify these problems early.
PFEIFFER: Oh, that’s interesting. Mucha gente está tan concentrada en los aspectos negativos de la IA y en el temor de que se vuelva deshonesta y cause caos. Pero usted está diciendo que aquí hay un recordatorio de que podría ayudarnos en esta situación.
ALPEROVITCH: Absolutely. Creo que va a hacer cosas increíbles para nosotros en muchas áreas diferentes, pero no hay duda de que cada tecnología tiene sus desventajas.
ALPEROVITCH: There is a lot of job security in cybersecurity.
ALPEROVITCH: Everyone should focus on resilience in their daily lives. Always have some cash on hand in case you are unable to withdraw your money from an ATM for a period of time. You know, stock up on some water, stock up on some groceries. I think those are prudent things in any scenario. Of course, there could be other catastrophes, weather-related, terrorism, etc., that you want to be prepared for. So that’s always a good thing to do.
PFEIFFER: Ese es Dmitri Alperovitch, experto en ciberseguridad y director ejecutivo de Silverado Policy Accelerator. Thank you.
Copyright © 2026 NPR. All rights reserved. Visite las páginas de términos de uso y permisos de nuestro sitio web en www.npr.org para obtener más información.
The accuracy and availability of transcripts for NPR programming may vary. The transcript text can also be reviewed to correct errors or match updates to the audio. Audio may be edited after its original broadcast or publication. The authorized record of NPR programming is the audio record.
Please note that NPR platforms and products also include certain programming produced by third parties; NPR no puede garantizar las transcripciones de ese contenido y algunas transcripciones pueden producirse utilizando herramientas automatizadas o de otro tipo.