AM Aalok Mehta Aalok Mehta Director of the Wadhwani AI Center at CSIS, where he leads research on AI policy.

“Cyber engineers and security engineers have to deal with a really rapid increase in discovery of vulnerabilities. They have to patch a lot more. They're under pressure to patch faster. And it's totally understandable that some of that patching doesn't go well and has led to some of these outages. It is quite possible that this represents the new normal.”

The AI Policy Podcast · Geopolitics & World Affairs · November 2026

“Cyber engineers and security engineers have to deal with a really rapid increase in discovery of vulnerabilities. They have to patch a lot more. They're under pressure to patch faster. And it's totally understandable that some of that patching doesn't go well and has led to some of these outages. It is quite possible that this represents the new normal.” — Aalok Mehta, The AI Policy Podcast

Context

Opening the episode, Aalok Mehta explains the week's wave of outages at Microsoft, Apple, Chase, the FAA and an NSA system. He says it isn't strictly an AI story, but AI-driven vulnerability discovery is forcing rushed patches.

Transcript

The AI Policy Podcast Around 02:41 into the episode
Nicole Herrera

Hello, everyone.

Chris Gorin

Hello, everyone. And yes, happy to be here. We're still on the air despite almost all internet services and flights systems being shut down. We are still on our mission to deliver you AI policy news every week.

Aalok Mehta

Yeah, so I'm sure a lot of the audience has heard the news this week that we've seen a ton of outages. Microsoft, Apple, Chase, the FAA had a massive outage that led to a significant number of planes being grounded for some time, a secure communication system at the National Security Agency. It's not strictly an AI-related story. There are lots of explanations for why this might be happening. But it is likely that some of this is just that cyber engineers and security engineers have to deal with a really rapid increase in discovery of vulnerabilities. They have to patch a lot more. They're under pressure to patch faster. And it's totally understandable that some of that patching sometimes doesn't go well and has led to some of these outages. And so it is quite possible that this represents the new normal and that we should prepare for more outages happening in the future on a regular basis.

Chris Gorin

Yeah, I also think it's interesting that anytime there is an outage now of any system, it's sort of like the first reaction is always assuming that it's an AI attack, right? Because that is, it's happening so frequently now. I think that's what a lot of people are assuming.

Aalok Mehta

I mean, there are a ton of new AI tools that help, you know, people who want to hack websites. And so I think it's quite possible. It's very likely that AI tools are helping whenever there's a hack. Some of these outages were definitely not hacks. They were things like a patch went wrong or an…

Nicole Herrera

Yeah. So today we have a ton to cover, specifically in the export control realm. That includes updates to the Jensen Huang saga and conclusive evidence that some of the latest AI models coming out of China were actually trained on export-controlled NVIDIA chips. So we're going to cover all that and…

Speaker names come from our own diarization. The position is estimated from where the line sits in the episode.

More from The AI Policy Podcast