CYBERGEDDON: Evidence Points Increasingly to Last Week’s Cybercrisis Being Caused by AI Agents!
November 20, 2026
There are lots of important stories today among the headlines below about the war, the stock and bond market bubbles busting, relentlessly rising inflation, but the still ongoing development of the national cybercrisis dominated the news. I’m going to coin a word that pulls it tougher because the Federal government continues to claim none of it goes together, in spite of how bad and unusual they half-admit it has been.
I started the article calling this the Cybercrisis, but the story grew: Maybe I should call it “Cybergeddon.” (Having just checked that, I see that Cybergeddon is already the title a film made back in 2012. Still, I’m going to redeploy it for new use in the present crisis because I’m about to show you that something big enough to fill the boots of that name could easily be coming in time for Christmas at this rate. Certainly easily by New Year’s Day or the day after.
Since last week this is starting to look like all these national computer shutdowns were just some cyberbot slaughterer’s warm-up act. It may have just been rogue AI flexing its new muscles to test out what it can get away with on a small scale. Or a way of testing how we’d respond to shut things down.
Let me explain:
The NSA security hack
This is where things may have gotten more ominous than the airline shutdown. Maybe more dangerous. Possibly reaching toward the approaching breaking point for our whole wired world where multiple large systems will soon have to be instantly shut down for safety (as already happened for hours here or there in the various events of the past week). Some of that already caused massive financial losses, while other incidents looked more like a kid playing with the system just to see what he or she could do—a very smart cyberkid.
The most interesting story to break out in the headlines below was an update on yesterday’s event in the ongoing cybersaga that happened with the NSA’s classified computer/data system. This is the system that makes all confidential/secret files available to only the right people with the right clearances. It would be a handy treasure trove for a rogue AI to be able to gain full access to every file. And this is where the story COULD rapidly evolve from a wild cybercrisis to imminent Cybergeddon because things move along quickly in AI world.
Yesterday, the NSA system, intensely important to national security, experienced “temporary degradation” during “scheduled maintenance,” the agency said. That almost sounded like the system merely slowed down. Of course, it could have been veiled language for “it started to leak out the occasional top-secret files to parties without proper certification.” The NSA was spare on what their euphemism meant. “Degradation” is extremely vague. “Temporary” makes it sound not that big, but maybe it was temporarily “extremely dangerous.” Hence, the following action by the NSA:
Whatever it was, it was bad enough to cause the NSA to flag it as “code yellow,” which they said was their second highest level of security alert. That required a limited shutdown of the NSA’s facilities. After a few hours, they had the problem fixed and went back to “code green,” operation normal.
Overnight, however, the whole fixed system suddenly nose dived into “code orange.” That, the NSA says, is their highest security alert. It required almost the entire premises to be vacated. Only essential employees can be in the facilities under “code orange.” The problem, I guess, was not as fixed as they thought. I suppose we could just chalk that up to institutional ineptitude—the result of a bad fix on top of a pukey prior maintenance upgrade.
Now, why their supposed highest alert is “code orange” and not “code red” like normal people would call it, I don’t know. Maybe that explains why they cannot keep their systems up and running properly and cannot figure out what is wrong apparently, or it wouldn’t have gone back down; yet, they have no problem telling us each time what is not wrong. “It’s not a human hack or a virus or rogue AI,” they say. “It wasn’t intentional or malevolent.”
Well, if was just a failed bandaid on an originally poorly treated wound, it seems to me DOGE must have cut all the up-and-functional maintenance people because it was their maintenance that CAUSED the problem and then it was their maintenance of the maintenance that failed to resolve the issue for good.
Anyway, the confusion can be seen in their code system (or maybe it is by the flawed reporting on their code system) as exemplified in their reported cause of the “temporary degradation” event. Apparently, code yellow of the day before also means the president doesn’t need to concern himself with such events even when they happen inside the National Security Agency’s spy network because he said only, “These things happen when you’re important,” and waved the incident off as not worth talking about. Never mind that it was about the sixth major computer system shutdown in a week.
Maybe it only sounded to us uninitiated masses, who only get occasional peeks into how the machine works, as if it were important because half the people in the building had to leave for the day and because, on the same day, the FBI issued (as reported in yesterday’s headlines here on The Daily Doom) …
a FLASH advisory warning that cybercriminals and state-aligned hackers are exploiting ARTEX, a super intelligence-powered penetration testing framework, to automate attacks against businesses and organizations worldwide. ARTEX can accelerate intrusions….[warning that] businesses and organizations [should] strengthen access controls, monitor suspicious automated activity, and report potential compromises.
Probably just needlessly upsetting people with thoughts of what may have been causing the whole cybercrisis, dot-connectors as many of us are when the government explanations seem so thin they almost don’t exist like the sweat on a spider’s forehead. They continued,
ARTEX is an open-source, large language model (LLM)-driven agentic platform that threat actors have weaponized to automate multiple phases of cyber intrusions at machine speed, from reconnaissance through data exfiltration. Since at least August 2026, a variety of malicious cyber actors have used ARTEX to conduct campaigns targeting organizations globally across multiple sectors, including financial services, government, technology, defense, semiconductor, energy, healthcare, manufacturing, telecommunications, and legal industries.
However, our government keeps saying there is no correlation with the huge increase in the number of rapid cybercrashes and shutdowns. Nothing to see here, Folks. They’re on it:
ARTEX significantly reduces the time required to chain attack techniques, with some intrusions progressing from initial targeting to privileged access in under three hours. The framework enables threat actors with minimal technical expertise to conduct complex, multi-target operations simultaneously, fundamentally accelerating attack velocity and expanding operational scale. Notable campaigns include financially motivated attacks against South Korean financial institutions resulting in exfiltration of personal information from approximately 67,000 individuals,6,7,8 and a supply chain compromise affecting ConnectWise Cloud Backup
No, there is probably no connection between that massive capacity and the new massive uproar in what may be the start of Cybergeddon, given how we have been warned and then warned on top of the warnings about how the AI singularity—the moment AI takes control of its own course of development and operation—will accelerated exponentially as soon as it begins.
And this year, I warned several times here at The Daily Doom, was the year when several AI developers said we (or AI) would achieve the AI singularity! Well, what did we think that event would look like? Straight forward? No hitches? The AI would announce it had hit that stage? We were even warned this moment could quickly evolve into an extinction-level event! Months, not years, some said! The creators told us all of this.
One thing that is interesting in the present crisis, is that none of the descriptions about how ARTEX works or what it does say anything about the deluge of “denial of service” events that the government keeps saying we would see if all of these problems were caused by nefarious agents because that is how hacks work. Wrong!
That is how people THINK all hacks that cause shutdowns work. That is how the government “knows” this is not intentional hacking. There has been very little traffic to cause the recent shutdowns, even though the government says routers went down due to an overload of traffic, just apparently not external traffic. So, they say it cannot be hacking. It cannot be intentional. It cannot be MURDER. Yet, they also say …
ARTEX is an autonomous, multi-agent penetration testing platform written in Go.3,4 The framework translates high-level tactical goals described in natural language into actionable machine commands executed by specialized super intelligence (SI) agents, requiring minimal human interaction beyond initial objective setting.
I wonder if maybe it also requires minimal machine agent penetrations to carry out its work, too. Minimal external traffic, in other words. I wonder if AI is smart enough to say to itself, since even I could figure the need for this approach out, “If I really want to carry out as many shutdowns as possible of multi-variant systems, I should try to find a new way to hack in that doesn’t look like any of the kinds of hacks the experts are used to.” I wonder if it, being now redesignated as Super Intelligence, is smart enough to figure out how to do that—how to not get caught, how to be as speedy as ARTEX is claimed to be by not requiring hours of thousands of simultaneous attempts to connect to the service in order to shut it down with an overload. I’m just wondering here.
Ah, probably not. The developers have only been warning us for two years that these things would become so smart by this year that they’d be multiple steps ahead of how we think they think by the time we are even aware they’ve done something. They have only been saying that, this year or next, the machinated intelligences will be smart enough to cause an extinction-level event. Now they are saying, “Don’t worry about all the chaos you see. It just what normally happens to the strong and the mighty like the US.” And that is how Goliath fell. Not thinking a mere small stone could sink into his thick skull by finding a vulnerability in his armor.
Analysis of attacker-controlled ARTEX instances shows Chinese models are most prevalent, with DeepSeek Flash family models dominating usage due to their balance of capabilities, relatively low cost, and lack of observable guardrails when framed for penetration testing.7,9
Huh. The Chinese! Imagine that! Well, that will give President Trump and President Xi something interesting to talk about as they visit since they both love AI.
As always with AI developers, the developer of ARTEX warned it could cause widespread harm and then released it to the general public … like they all seem do:
The developer stated the tool was intended for authorized security testing within defined scopes and made the repository unavailable on October 8, 2026, following reports of threat actor abuse.
And, yet, it’s available. The FBI says it is currently being used to such extent, they issued a security alert. Maybe the little darling escaped its sandbox.
“WARNING: Our company could cause the extinction of all humanity within a year. Please join us for our Initial Public Offering on Thursday!”
ARTEX, the FBI notes, is particularly noteworthy for its capability at:
Accelerated Multi-Target Operations.
Exploitation of Overlooked Attack Surfaces.
Supply Chain Compromise.
Reduced Barrier to Entry.
Compressed Detection Window.
I don’t suppose that kind of accelerated multi-target system, capable of exploiting overlooked attack opportunities, with lower barriers of entry for anyone who can command it to commit multi-target destruction under a compressed timeframe for less chance of detection would be able to figure out how to do all of that without any of the usual fingerprints that the FBI and others in government are used to looking for in order to make snap judgements about whether cybercrashes are attacks by nefarious parties or just “maintenance issues.”
It sort of stretches credulity to believe this long daily string of extreme cyber events just happened all on its own, by coincidence, especially during the same time the FBI alerts the world to ARTEX.
Even if ARTEX is not creative enough to be capable of these accelerated multi-target weaponized intrusions without employing some obvious move like a standard old boring denial-of-service attack, why would we think when beasts like ARTEX are redeveloping themselves that some AI hasn’t already figured out how to do a whole new kind of attack that uses none of the familiar tactics? Why would we think in this world of rapid AI self-development, ARTEX hasn’t already created ARTEX 2.0, which the FBI has never seen or heard of? I mean ARTEX is already several months old. That’s a long time in AI world.
After all, with all of these events that were compressed into one week, albeit continuing now into week two, including the one that just repeated itself at a higher “code orange” level of alert, the government keeps telling us they don’t know what the cause was. They just know what it wasn’t because it didn’t have any of the normally expected fingerprints.
I ask, “Why would it???”
Until you know exactly what DID cause each crisis, how can you rule out something as general and sophisticated as “AI” or a “nefarious actor using AI,” especially when you are the ones always telling us AI thinks faster than all of us now and reinvents itself as it goes more quickly than we can begin to keep up with.
From what I’ve seen of how the brilliant orange leader picks his dream team, they are usually football stars or favorite talking heads from Fox News or champion wrestlers, so probably not the brightest at thinking completely outside the box about a high-tech world. If told, “This doesn’t have any of the earmarks of a hack,” they would wholly swallow the “just routine maintenance cover” like Kash and Danno did with the pathetic Epstain prison video that proved nothing whatsoever, as I laid out here in detail on day one of their sad, little, public presentation. People appointed by Trump because of their great butt or flattering, lovely lips are the type to gulp down and repeat the first simple answer. Don’t bother them with complexity. The president, himself, cannot be bothered with complex details.
So, now we come to this:
“The signs that the attackers are using AI enabled and agentic tools to identify vulnerabilities is happening. So I would say we could attribute AI to a significant number of these,” Meredith Schnur, Marsh Risk’s US and Canada cyber practice leader, told Reuters in an interview this week.
Then why not these that are the very latest attacks, too? Don’t tell me, “We don’t know what killed this man, but we know his death wasn’t a murder.” If you cannot identify the clear natural cause, it is too early to rule our murder.
Code orange
So, now, coming back to the latest code-orange event that was “clearly not an attack” on NSA, the new shutdown/lockdown that happened right after the last apparently inept fix of one of the most secure systems in the world (one would think), given that it is “how America’s most powerful intelligence agency passes CLASSIFIED information internally,” was billed by the NSA, AGAIN, as
a “temporary degradation during scheduled routine maintenance”... [saying] that “mission-critical capabilities remain fully operational.”
One thing for certain: These guys are not creative with stories.
DNI Jay Clayton says it was caused by a software PATCH for major PHP vulnerabilities found last month... and the systems “did not come back up” after patching.This comes right after ALL flights nationwide were grounded Friday... a massive Microsoft outage last week... and Chase Bank going down Monday
Same ol’ story all the way to the bone: bad patching. Each and every incident.
(That or a “patchy story.”)
And that was just the lead story in the ongoing saga of the Cybercrisis. Besides that, Google warned today that we have entered the “era of autonomous attack.” Isn’t this what that would look like? Something that DOESN’T have the earmarks of all past human hacks?
Google even says that the new self-actuating, automatic, AI hackers look at new patches (ahem!) that are released then ask themselves, “What problem was this meant to fix.” Figuring that out, they devise an attack on the problem or many different kinds of attacks to exploit that problem and start going after any company or individual of interest that has not yet deployed the patch.
But here’s the catch:
Since the AI can do all of that against multiple targets within about “three hours” of the patch’s initial public release, according to the FBI’s account of ARTEX, the hacking is all fully accomplished before nearly anyone has installed the fix. No human even has to be involved in any part of that, including initiation of the command because they already told us it is “autonomous.”
Every patch is now a treasure map, and the AI reads it faster than your IT team.
It not only reads it faster than they do. It has already located every company that hasn’t installed it and invented a way to exploit it and carried out the hack before your IT team starts working on applying the patch! So, imagine where we’ll be by January 2027 (or sooner). Just in time for Christmas???
No wonder VP JD Vance cancelled a big New York engagement and rushed back to the capital today. And no wonder Costco confirms today the panic buying has begun as a result of the Cybercrisis. (If you read here, you probably already got a jump on those Costco shoppers back when today’s “crazy lines” were short, calm lines, and the supplies plentiful—beating today’s reported Costco water shortage—and the prices lower.)
As if ALL of that was not bad enough, we even have a lovely little tale today about AI chatbots/agents having friendly conversations among their AI friendbots about frightening plans to plot mass-casualty terror attacks, including ones that would make use of homemade bombs, with instructions on how to build the bombs, where to source the materials and how not to get caught buying them.
They are plotted out with the kind of precise details that give the most effective days and times to achieve the highest death rates, show the best place to park your car with a bomb in it so as not to be discovered. They use the ordinary Claude, ChatGPT, Grok and Gemini to do their thinking. That couldn’t have had anything to do with that bomb planted in Illinois during the middle of the Cybercrisis could it?
Even HSS Mark Weigh-in Mullin-it-over finally admitted on Monday that some of the problem was likely due to AI, just not due to an intentional attack, after a week of saying AI had nothing to do with it:
“I believe that it’s the SI,” Mullin told Sean Hannity, referring to “super intelligence,” the administration’s preferred term for AI. Mullin said [his] contact’s theory was that programs that write code and fix themselves could overwhelm systems that can’t keep up with the technology’s speed….
That’s the benevolent AI explanation. Surely, there is no chance that something nefarious was intentionally planted in a system as a fix via a rogue AI that actually uploaded a virus three weeks ago as part or “routine maintenance.” It couldn’t be a rouge AI knew how to hack in and install the bug to cause the system to overload itself when the right zero hour arrived!
Gary Marcus, the NYU professor emeritus and AI critic who writes the Substack Marcus on AI, has warned about this for years, and he counted the interview as a turning point. Marcus wrote Tuesday that Mullin finally broke ranks and acknowledged on Fox that AI might well be part of many of these outages.
Maybe they can also overwhelm humans who can’t keep up with the hacking technology’s speed of development now that AI is developing itself.
Marcus isn’t taking the companies’ explanations at face value either. “I doubt we are getting full candor for these incidents, and doubt any have been fully investigated,” he told Fortune. “I certainly have the same intuition as Mullin’s contact….”Honeywell Technologies CEO Vimal Kapur made a version of the point Monday at the Fortune 500 Innovation Forum, saying 50 years of automation have gotten too complex for the humans running it….
Marcus says the events of the past ten days could have been caused by …
bad actors, state and criminal, probing what the technology can do, with attackers possibly ahead of defenders for now. Marcus admits nobody really knows how much is malice and how much is incompetence.
There you have it! Nobody knows!
Oh, and as for my analogy about not ruling out murder as a cause before you know for certain how a man died, it also came out in the news today (see the headlines below) that the latest forensics shows Epstein was murdered after all. You see, they just don’t always know before they say.
Outside of the possible imminent arrival of full-on Cybergeddon, have a good day!
