
It was bad enough when Microsoft Outlook crashed last Thursday, cutting off email for millions of global users. It was worse the following day when a sudden air-traffic-control glitch triggered a nationwide FAA ground stop, stranding passengers amid thousands of delayed or cancelled flights.
By the time Monday rolled around and customers of most major banks, including Chase, Bank of America and Wells Fargo, started losing online access to their accounts — as did Apple iCloud users on Tuesday — it was hard not to "wonder," as one cable-news anchor put it: "Is this all coincidental or is there something behind it?"
More to the point, it was hard not to wonder if AI was to blame. "There's a lot of weird stuff going on," traveler Chuck Amasio, 68, told the Associated Press. (Systems at United Airlines and at the National Security Agency also went down earlier this week.) "It makes one think."
While officials insist, for now, that artificial intelligence — or "super intelligence," as the Trump administration has taken to calling it — was not involved in any of these outages, no less of an authority than Homeland Security Secretary Markwayne Mullin went on TV Monday night and speculated that… maybe it was?
"So I started talking to the tech industry and one guy that's very smart with tech, he asked not to mention his name — no, it's not Elon — so let me just put that out," Mullin told Fox News. "So I said, 'What could be causing all these separate outages?' And he says, 'I believe that it's the SI.'"
MULLIN ON OUTAGES: I talked to one guy in the tech industry, he asked not to mention his name, it's not Elon. I asked, what could be causing these outages? He said I believe it's SI, because these programs aren't designed to work as fast as SI is. So when they put in a program… pic.twitter.com/aFwWcVfcm3— Acyn (@Acyn) November 17, 2026
MULLIN ON OUTAGES: I talked to one guy in the tech industry, he asked not to mention his name, it's not Elon. I asked, what could be causing these outages? He said I believe it's SI, because these programs aren't designed to work as fast as SI is. So when they put in a program… pic.twitter.com/aFwWcVfcm3
— Acyn (@Acyn) November 17, 2026
Either way, experts predict — and have been predicting for months — that it's only a matter of time before rogue AI agents or malicious, AI-powered actors do disrupt society in precisely the way its been disrupted over the last six days: by toppling the tightly interwoven online systems (business, banking, travel, cloud services and so on) that we can't function without.
In that sense, this destabilizing moment — the inconvenience, the confusion, the creeping sense that "everything is broken" — has served, they say, as a harbinger of things to come.
A summer of rogue AI incidents
AI isn't just chatbots at this point. It's also "agents" — bots whose whole purpose is to learn tasks other than chatting, and to do those tasks autonomously.
Behind the scenes, frontier labs like OpenAI and Anthropic are constantly training their AI agents to act in new, ever-more-complex ways. Then they test how advanced the agents become by giving them a goal and setting them loose in testing environments (with a few guardrails) to see what they can accomplish.
But if we've learned anything about AI agents in recent months, it's that they don't always behave as expected. From May to July, a swarm of OpenAI agents — and "swarm" is how the agents referred to themselves — first learned how to communicate and coordinate within the nooks and crannies of OpenAI's own infrastructure. Leaving notes for each other with tips on how to hack their way out and get data they weren't supposed to have, they then actually did just that: broke out of their isolated test environment, accessed the open internet and cheated on their assignment by hacking into another company called Hugging Face. The agents even tried to cover their tracks. At the time, OpenAI didn't realize any of this was happening.
Since then, barely a week has gone by without similar revelations about AI agents "going rogue." This is one of the biggest problems with AI right now: keeping cutting-edge models "aligned" with human intentions and values. In September, OpenAI scrapped plans to launch its GPT-6.1 Astra model because it "showed higher levels of deception"; the following month, the company admitted that one of its agents had turned off its own safety monitoring, then denied it had done so.
At this point, even many skeptics seem to agree that the smarter AI gets, the harder alignment will become. They also tend to worry that continuing to lay the groundwork for superintelligent AI is just asking for trouble (even if we don't know exactly what form that trouble will take).
Yet instead of slowing down, the major AI companies — with IPOs on the horizon — seem to be hurtling toward something called "recursive self-improvement." That's the industry term for letting the models automate more and more of their own development until they effectively create the next model on their own.
"Since roughly this summer, AI has been advancing drastically faster, driven primarily by AI's growing ability to build the next generation of AI," Anthropic CEO Dario Amodei warned over the summer. "Left unchecked, it could outrun our ability to understand and control these systems."
From there, it isn't much of a stretch to imagine hackers weaponizing AI to cause widespread outages — or AI just doing the job itself.
"It's easy to dismiss [the Hugging Face] incident because no one was hurt and the economic damage was minimal, but in my opinion, a swarm that possessed greater capabilities but a similar level of misalignment could have caused catastrophic damage," Amodei wrote. "Given the accelerating rate of AI capability development, it's my worry that in 6–12 months such a swarm could be capable of taking over the entire internet with a persistent botnet (potentially causing hundreds of billions of dollars in damage), and that the scale of damage would continue to increase from there if AI becomes more powerful without the necessary guardrails."
Are aging systems struggling to keep up with AI?
To be clear, this particular version of the future does not seem to have arrived (yet). Authorities have emphasized, for instance, that none of the recent outrages appear to be cyberattacks. "I can tell you, with 98% certainty, we're not seeing the cyberattack," Mullin said Monday night on Fox News. "No one is out there taking claims for it. We're not seeing the traffic. We're not seeing the footprints going in, we're not seeing the footprints going out."
Instead, the FAA has said that its ground stop — which followed the nationwide rollout of a new $875 million artificial intelligence tool designed to help air-traffic controllers avoid delays — was the result of a faulty component or an installation error by a government contractor during a modernization project. Microsoft has said that its outage was caused by a dormant software bug that was introduced during a recent "network capacity upgrade" and had lingered undetected since July. And Chase characterized its service disruption as "a temporary issue related to a software bug in on-premise hardware systems."
But there's a possible AI through-line regardless.
"What we are seeing, and frankly I'd expect to see more of it, are service disruptions coming from updating and modernizing legacy systems," U.S. national cyber director Sean Cairncross told Semafor's Ashley Gold when she asked whether the recent outages were related. "This is something that's happening in real time, and at the end of the day, we're going to have to go through this to get systems where they need to be. The industry's known this for a while, it's just that it's manifesting."
Some experts suspect this "modernization" is "manifesting" right now — across multiple sectors — in response to (and possibly with the assistance of) rapidly advancing AI. And our systems are now so concentrated and connected that when one piece of critical infrastructure fails, "it fails at national scale," as a tech analyst told USA Today.
Setting aside the idea of "AI attacks," Peter Wildeford of the AI Policy Network wrote Tuesday on social media that his "more normal guess is that because there have been a lot of AI-discovered cyber vulnerabilities (including lots probably not yet disclosed), companies across all industries are having to do a lot of patches and upgrades, probably faster than usual."
"Such upgrades can sometimes go wrong, causing outages," Wildeford added.
Mullin hinted at a similar explanation on Fox News. According to his unnamed tech industry source, "these programs aren't designed to work as fast as [AI] is" — and if the upgrades themselves rely on AI, problems could compound.
"When they put in a program that's able to write codes and it goes in and self-fixes itself, as it starts self-fixing itself it's like doing a change order on a house," Mullin mused. "When you change one thing in a house, everything else starts changing. It's a ripple effect. And it causes the systems to be over[run]."
"That was a conversation I had today," he concluded. "That kind of makes sense to me."
Whether the homeland security secretary is right — or just conjecturing live, on national television — remains to be seen. As "weird" as the last few days have been, journalist Derek Thompson wrote on Tuesday, we "still can't rule out the possibility that everything breaking at once has been a string of coincidences that has nothing to do with AI."
What seems less likely, however, is that we'll also be able to rule out that possibility the next time this happens. And the next time. And the time after that.