Welcome to Doomsday Scenario, my regular column on national security, geopolitics, history, and—unfortunately—the fight for democracy in the Trump era. I hope if you’re coming to this online, you’ll consider subscribing right here. It’s easy—and free:

I originally started this newsletter — and chose its name, “Doomsday Scenario” — to be tongue-in-cheek. I hoped/assumed that I would examine the question “are things really as bad as they seem?” and generally conclude: “No, they aren’t. This isn’t the doomsday scenario.” Of course, then Trump happened, and I’ve spent much of these last two years writing about how often things have been worse than they might seem.

Today, I wanted to write about something that definitely isn’t as bad as it seems: Don’t be worried about all these tech outages and oddities over the last week — or, at least, don’t be worried about most of them!

It’s easy to see why the last week has everyone worked out: There was a massive Microsoft outage on Thursday, a nationwide groundstop by the FAA on Friday that disrupted travel for hundreds of thousands of people, and then a further wave of website outages on Sunday, Monday, and Tuesday, that affected variously banks, Apple, and United Airlines, among others.

There are two interesting things happening right now — first, the actual incidents, and second, what they’re showing about the state of our government.

First, the tech.

There is something interesting happening in tech right now, but it’s actually about the battle to protect against nefarious actors — not the nefarious actors themselves.

From everything we know so far, there is every reason to believe that there is no nefarious actor behind this. I’ve actually spent the last two days in Washington, D.C., at the Aspen Cyber Summit, in my Aspen Institute cyber policy hat, and talked first-hand to most of the people in the country who have the best understanding of what’s happening.

Their consensus is clear: This is not an attack on the United States.

Instead what we’re living through right now is a moment when the defenders — the people charged with keeping networks and systems safe, and patching software vulnerabilities when they’re found — are going through one of the most intense periods in the history of computing.

The new powerful LLMs by frontier AI companies — like Google, OpenAI, and Anthropic — are doing work in minutes and hours that would have taken the world’s most sophisticated hackers weeks and months. That means companies are identifying new system and software vulnerabilities at a blindingly fast pace right now — which means, in turn, that the defenders are installing and updating software patches far more quickly than normal.

Normally software patching — particularly in complex systems or critical infrastructure, like banking or, say, Microsoft Outlook — is slow, and involves an enormous amount of careful testing to double-check that the patch doesn’t actually accidentally break anything and end up doing more harm than good. Patches often roll-out slowly and might fix multiple vulnerabilities at once — think more like a regularly scheduled weekly, monthly, or quarterly doctor’s appointment for your software to deal with various ailments at once.

That’s not happening right now, or at least it’s not happening as robustly as normal, because of both (a) the volume of vulnerabilities that are being discovered and (b) the speed with which malicious actors, using the same software tools, are also discovering and exploiting those same vulnerabilities.

Right now, though, the world’s software is more headed to urgent care — or maybe even the ER.

These new advanced AI systems are finding vulnerabilities in existing legacy tech systems quickly — and they’re being exploited quickly, so companies are rushing the patches into service as soon as they’re identified and ready. What we’re seeing in all these outages are basically patches breaking legacy systems — problems that would normally be caught in careful behind-the-scenes testings, but now are rolling straight out to live systems.

Patches are happening fast enough that some systems are basically having new patches break old patches too.

I’m obviously oversimplifying some of this, but the bottom line is that you’re seeing software problems that normally would be caught in slow, careful, deliberate testing instead happen on live systems and outages ripple through to consumers.

There are several addition layers to this — a lot of companies have deployed new AI-written code into systems in recent months that hasn’t necessarily proved as smooth as originally believed, and you’ve seen mass tech layoffs and retirements and reorgs over recent years remove a lot of the institutional memory for legacy systems inside these companies, so they’re not necessarily able to fix them as quickly as they might have in the past.

(The FAA outage appears a bit different — but still similar in that it’s a struggle to update old outdated systems to more modern ones. It’s what the industry calls “tech debt” — how we’re paying the price for old systems.)

Anne explained, “The volume [of vulnerabilities and patches] has fundamentally changed,” citing that Microsoft, for instance, has in the last month patched ten times the number of vulnerabilities that it did in all of 2025.

Speaking at the summit too, Crowdstrike’s Drew Bagley explained that they’ve seen their “time in the wild,” e.g., the time that companies have tracked from understanding a vulnerability to seeing it exploited by a bad actor collapse from a traditional time window of “days” to, over the last year, first “hours” and then “minutes” and now even negative, e.g., exploits are happening even before defenders understand the vulnerability. That’s a major sea change and one that will likely mean that these outages — once rare — might become a more regular feature of the coming weeks and months, at least until a new stasis or equilibrium is established between attackers and defenders.

The White House clown show. (Photo by Alex Wong/Getty Images)

The second strange thing that’s happening amid all of this, though, is the failure of Transportation Secretary Sean Duffy or DHS Secretary Markwayne Mullin to project any calm or competence amid these problems.

Their statements have been disjointed, combative, and Mullin has now gotten in several days of controversy over how he flew on Saturday, amid the post-FAA outage chaos, to sit in a box seat at an LSU football game with a major DHS contractor. (The fact that this controversy started at the New York Post shows that the knives are out for Mullin inside Trumpworld!)

We’ve now seen instead Marco Rubio, in Asia on the President’s strange tour, take the role out front, sending a different message to try to calm the waters here.

POLITICO reports that Mullin is now on a “TV time-out,” based on orders from Stephen Miller.

The failure of Duffy and Mullin to communicate in this moment is another sign of how problematic it is that Cabinet secretaries consider their primary audience for public statements to be Donald Trump — and not the American people.

The public needs to have confidence in their leaders when crises arrive — and neither the reality TV star-turned-Transportation Secretary nor the plumber-turned-DHS secretary have been able to project any calm in this moment.

That to me is almost a bigger problem than the outages themselves.

If you want to read more about any of this, I’d direct you to two Very Smart People: My friend Jeff Greene, who served as one of the nation’s top cybersecurity officials, about the website outages, and then James Fallows, who is as loyal readers know always the person I turn to for aviation coverage.

Fallows, in particular, has what might be the best-timed New York Times Sunday Magazine cover story ever, which ran last Sunday two days after the FAA outage, about whether air traffic is actually getting less safe.

Thanks for reading!

GMG

PS: If you’ve found this useful, I hope you’ll consider subscribing and sharing this newsletter with a few friends: