Also, I don't think it's that "AI code is inferior" but rather devs got sloppy and no longer review code. Not to say that everyone's lazy but me---a lot of systemic/environmental factors could be the reason.
People were making memes about the GitHub unicorn a decade ago.
This is a systemic issue that's driven by AI but the responsibility lies with leadership who are still operating on the move fast and break things mentality. With everybody swinging for the fences and the median blast radius increasing and the checks/roadbumps eliminated you end up with a lot of incidents because the total number of merges and the size of the diff has increased by quite a bit per sprint.
shit programmers have always been a thing, these same shit programmers now putting output from claude straight to production
on top of this there are literally PMs getting involved and doing their own PRs, see vscode recently
Chernobyl was caused by a guy flipping a switch, Challenger blew up because of a faulty O-ring, etc.
No, just no...Chernobyl was caused by the staff doing an experiment that to put it mildly was wildly risky. They disabled safety systems with wrenches (meaning they couldn't easily be reenabled). Challenger blew up because staff at NASA ignored warnings from the people who made the O-rings. Saying these were "straws" is completely irresponsible. Attitudes like this are what cause outrages. Also, unless you are being dishonest, this is the worst string of outrages in the history of tech. There is a cause. It can be found. And unless you think in these terms, you have no business working with a production system in any way.
I also wonder if Trump's administrations' spokespeople even realize they have a massive credibility problem, and that if they want people like me to believe anything then I have to hear it from a mid-level functionary, not a political appointee.
You say whatever makes your stocks go up and now down, and if you don't, you are stupid because it's normalized now, there are no repercussions, and if you don't someone else will.
>Do you really believe
Thanks! Conspiratorial thinking hadn't annihilated my last brain cell yet.[1] https://apnews.com/article/faa-smart-ai-outage-08af345749c81...
[2] https://www.faa.gov/newsroom/nationwide-ground-stop-question...
It's caused my AI, it doesn't really matter who's fault it is
Vibe coding, vibe reviewing, vibe fixing. No one has a clue how software works anymore.
Somehow I knew I would find something like this on this comment section.
Basically, blaming AI slop supports a lot of pre-existing beliefs, while blaming AI hackers does not.
- Physical branches closed or unresponsive.
- Online banking down.
This exact same thing happened in Europe with Revolut late last month. It was resolved in about 12 hours: https://www.revolut.com/blog/post/20octoutagetechnicalovervi...
In that case, it was a Google Cloud outage. Something similar (database outage) could be the culprit here.
Things like food, water, basic sanitary items, would be bartered for.
Basically, these guys: https://www.nytimes.com/2025/11/22/us/texas-men-haiti-island...
Anyone known to be a prepper is a really great target in an apocalypse. One-stop shopping and typically doesn't work well with others!
All hat, no cattle.
.223, while we're on the subject, is used in the AR-15.
-Deng Xiaoping (paraphrased)
There are definitely issues with that, but it does work in the sense that people can buy and sell things and survive, provided the government is competent enough to provide good quality notes and coins and resists the temptation to print too many.
After all it’s how western societies worked throughout human history until relatively recently
On the other hand stablecoins such as USDC can be sent over the internet on blockchains such as Tempo, Solana, XRP and Stellar worldwide, 24/7 in less than 5 mins and with extremely low fees.
Roman citizen wondering why the aqueduct isn’t working kind of thing.
When banks started moving to computerized systems, maybe a bank would be unable to process something... but not multiple banks. I can't recall any "outages" that affected banks nationwide for any period of time. Walk into your local bank and find out they can't do something because their big system at the mothership is down? Perhaps.
Consolidation likely plays a part in this as well. I can't speak to the FAA/ATC systems 50 years ago, I have no recollection of what they were like then, but the only nationwide system I can think of around that era was Ma Bell. And I don't honestly recall any nationwide outages back then. Regional, sure. You might have a problem with part of the country getting calls if there was some kind of emergency... (e.g., everybody trying to call their grandmas in Kansas after reports of a tornado) but I don't remember there being any nation-wide problems like that.
We had to count on actual snow for snowdays. :-)
There was this one in 1990 [1], but it's notable because the system was otherwise very reliable. Also, this report says only 50% of call attempts failed, presumably most people would try their call a few times and likely get through.
[1] https://users.csc.calpoly.edu/~jdalbey/SWE/Papers/att_collap...
In 2006, Amazon had just launched this new storage service called "S3." It would have struck people in 2006 as bizarre to think that AWS (and Cloudflare, and GCP, and friends) would one day be so crucial to the internet that a bug there could take down dozens or hundreds of companies.
When I hear people talk about "we can't do big things anymore" in the US, they are usually referring to mid 20th century stuff a la https://patrickcollison.com/fast, not stuff that we did 10 years ago.
Not the things that are breaking, no. The FAA was doing ATC a decade ago. We had email and cloud servers a decade ago. We had bank websites a decade ago. The scale of those things today is perhaps larger than a decade ago, but not enormously larger; indeed, the increase in scale in those things is much smaller than the increase in computing power in the world from a decade ago. So these things should have gotten more reliable--but they haven't.
we made these systems simpler to maintain, to reduce the amount of skill we need to pay for to maintain them. And that does reduce the maintenance costs, at the cost of more compute. but who cares about compute costs where compute per unit of money doubles every 18 months, amirite?
unfortunately it also increases the competence required to change these systems without degrading their performance and usability.
since no one wants to pay that extra cost either, we let the systems degrade a little bit at every change instead, hoping they degrade slower than hardware improves.
that bet was bound to lose eventually, because of laws of physics, and because that degradation pattern affects hardware too.
and the double whammy is - with little demand for expertise - the supply is effectively gone too. so now that the walls have closed in, there is no way out.
[1]https://www.fintechfutures.com/baas/bt-fault-hits-uk-banks-d...
[2]https://abcnews.com/US/hackers-block-us-banks-online-service...
But for everything else you need, food water shelter healthcare, you need to submit to europay visa mastercard
They are entirely wrong.
None of these outages, as far as we know, have a common cause. They just happened by coincidence inside the same news cycle.
Nor is there any evidence to blame them on AI, or tech layoffs, or hackers from Iran. It's more likely something boring and technical that doesn't fit into anyone's narrative.
1. Does there seem to be a larger concentration of outages recently? (I would say yes.)
2. Is there actually a larger concentration of outages recently, or does it only seem that way because of an increased focus on this? (I would say that there is really a larger concentration, but I'm less certain about this.)
3. If there is actually a larger concentration of outages recently, is that because of AI? (This is the point at which I become pretty unsure either way.)
4. If it is because of AI, how exactly is AI causing this? (I think more widespread vulnerability discovery / exploitation is more likely than "everyone is vibecoding everything now and that's screwing everything up", but I'm not entirely sure.)
5. Is this just transitional growing pains, or is this a new normal, or is this the beginning phase of a pattern that is only going to accelerate? (I don't know the answer to this one at all! I hope it's the first thing, but I wish I were more confident.)
Either way, now that AI is more capable, you can really mess stuff up at scale.
One way or another, they've tried to avoid addressing tech debt, avoid hiring and retaining good people, and have more or less been kicking the can down the road / running on the edge without knowing it.
Or else they did embrace AI-driven "move fast and break things" approaches, and so they broke things.
They flip-flop between maximizing profits and draining operational divisions of any resources, and hoping to sprinkle a bunch of IBM magic pixie dust all over their infrastructure to solve all their problems without needing to understand any of it. You rarely find one that strives to understand problems deeply and invests in long-term planning and maintenance before a crisis hits.
The result is old systems that eventually break, system upgrades that go bad because nobody understands them, security holes that someone might have hacked into, etc. All the same basic root cause at the top, which is that most of our leadership class is full of incompetent short-sighted narcissists. And it "works" because most of the time it pumps the stock due to lower costs, and we only see the fraction of times that it causes headline-grabbing catastrophes. On a long enough timeline, though, they'll happily drive our entire society off a cliff--and we're getting closer and closer to that (at least politically).
"As far as we know" just indicates a lack of data, not evidence that they're not related. This is especially relevant for a political administration whose response to COVID was to bury their head in the sand by hiding data. Also: A public comment of "AI is hacking all our shit" wouldn't exactly calm the public.
It's fine to say, "I'll wait until we know more", but in the meantime AI progress continues and the sheer volume of cyber vulnerabilities it's uncovering is undeniable. It's not surprising that some state-sponsored bad actors would be use this as an opportunity to infiltrate too...
In the absence of data, you are assigning blame to things you already hate, like AI or the administration. Motivated reasoning is highly prone to error, and it's a form of stupid-brained thinking.
It's your job to provide evidence they are related, and you have nothing except speculation.
Microsoft on the other hand, has proven that it will never be reliable.
You can argue that this is a result of something outside of their control, and you would not be wrong, but if the point here is about centralization rather than simply a particular company's operational procedures, then direct strikes on datacenters should be part of that conversation.
I was an Amazon red badge in the EC2 org, fwiw, and do think AWS has excellent operational culture, far beyond what I saw while working in the M365 org, and beyond what I've seen at other tech companies and hedge funds where I've worked. But that doesn't obviate issues with centralization.
If Bank of America's central datacenter (cloud or on-prem) gets hit by a cluster bomb, taking out banking for ~half the US, you can be damn sure that very smart people are going to be working 24/7 until it's resolved.
If you have all your savings stored with a local credit union, say Seattle Credit Union which has 50,000 customers, and they get hit with a ransomware attack... well, good luck. You (probably) won't lose any actual money thanks to FDIC rules, but it might be weeks or months before their services are back online.
Similar to self-hosting vs cloud, you get a bit of a paradox where it's overall better for society if there's no interdependencies so a SPoF outage doesn't impact everyone, but things are much worse for an individual in the event a localized outage does occur due to economies of scale, or lack thereof in this case.
Its pros/cons are known and gaps can be addressed with 3p software. You just keep on old outdated, but proven version of software and keep the perimeter locked tightly.
its the azure cloud that is the wildcard.
Don't forget that MS has the advantage of non-English languages support that many alternatives lack.
I feel like it is very unknown, there is no documentation and everyone relies on comments on random forums that are now 21 years old which tells to set undocumented registry keys to undocumented values.
I prefer to ask a different question. Will all this AI development delivering us amazing new software? Are we just going to get 1000 free clones of Adobe Photoshop or can we expect some really exciting groundbreaking new tools? Perhaps all the really clever use of AI development is in the pursuit of better AI and if that’s the case, why?
As to why, well economy must grow, no matter what.
Unfortunately it will take many more outages for folks to realize that serious engineering talent has great value in the AI age. It's their judgement that helps speeding headfirst into building unreliable unmaintainable and difficult to extend systems.
Some other thoughts:
* Will the Internet become balkanised?
* These smaller networks will require a lot of personal information to even gain access.
* The return of the Sneakernet?
/s
not sure if vibecoding is a cause or consequence, but it’s related, it requires as an axiom to stop caring about internals. And it seems to require disregard for quality too.
i'm all for experimentation and adding new tools when they're shown to actually help, but it seems like large software companies (especially banks and e-commerce) just don't care about quality anymore.
All of this is also downstream of what I think is even more important, which is what's better for society in general. Obviously low quality software is not better for society than high quality in a vaccuum, but I also don't think there's a single obvious objective answer for whether the influence of software on society is a net positive or negative today compared to, for example, when computers were all mainframes and all software was produced by people paid to write it. There are a lot of differences in society today compared to back then, and software certainly has some role in how we ended up here, but trying to attribute exactly what and how much is not an exact science.
This is a lot of words to basically say: if you're confident that this is a problem, I think it's important to state exactly why, since I think the interest parts of the discussion are in how you've concluded that.
Commonly asserted but not true. Most programmers are not just letting the model take the wheel.
> it'll probably be part of the solution.
LLMs can't solve the slop problem they themselves are creating.
Uptime and quality of software is about 10000x better than it was 10 years ago. Things just work all the time now.
Extraordinary claims require extraordinary evidence. You're being the 'trust me bro' in this situation.
I told him it's just like how he keeps telling his patients to watch what they eat, drink and smoke less, and get some exercise, but they don't, and then they have a heart attack and wonder why.
But I think there's an increasing consensus among "normies" like your coworkers who see these as related. They don't know why but assuring people they are unrelated fails in messaging terms when it feels related.
I think this will be an increasing problem where so much more code is not only written by AI but reviewed by AI. Things will slip through and companies won't be punished for these outages in terms of public perception (because it'll happen to everyone) or by the government. We just don't have serious consequences for corporate malfeasance or gross negligence. Like AFAICT the Sackler family should all die penniless in 6x8' grey prison cells.
Unfortunately we live in a country where the Sacklers get to stay billionaires while 1 year olds are shot dead at the behest of Walmart over a pack of diapers that weren't even stolen... with zero consequences [1].
It feels like this is just the next level of entshittification.
[1]: https://www.nbcnews.com/news/us-news/mississippi-ag-says-pol...
People "vibing" everything is just another component of this, not the root cause. Using AI tooling is just the new technique for what used to be called phoning it in, or going through the motions, or faking it until you make it.
Unfortunately, it's nearly always been an effective tactic for a selfish few. It's problematic when we don't have an equilibrium that keeps it in the minority...
(The same thing happened with separate company's problems becoming "Big Tech." It didn't help.)
Snarky, but there's a point: Abstraction leads to viewing different things as the same thing. They are - at one level of abstraction. And they aren't at other levels of abstraction. It's not a problem to have both viewpoints.
The trick is that, when someone blends things together, to not lose the fact that they are also separate (at other levels of abstraction).
So I have no problem saying that "optimizing for efficiency and speed" is the common cause of all of these problems, and saying that they all have different (immediate) causes. Both can be true.
I'm not sure "optimizing for speed and efficiency" can count as a common cause, though, because it's not a single event. It's more like a set of events that have something in common.
The main incentive for any nation state to avoid attribution to direct and material harm to many Americans is that they risk awakening American bloodlust. Americans are unhappy about the current Iran situation and wish it didn't start and that it would end already, but sufficient Iranian escalation would turn that around real quick and suddenly there are boots on the ground among other things.
Saying it's not related applies an assumption of what level of specificity you are thinking about. And that's how you quickly de-sync with other people, especially "lay people" who look at you funny when you claim there's just no way these things are related when it's hard to believe that based on experience alone.
As other comments suggest, it probably is related to an extent. Related by complexity of our Internet-based systems, related by age, related maybe even by enshittification of expertise, etc.
The further up the tree you find these relationships, the less specifically actionable they are (you cannot CVE enshittification), and the more they are about "what are the overall social/cultural/human issues we're grappling with in this era?"
Sure, focusing on individual small causes is helpful too, and it's a lot harder to do the higher and higher level system stuff. But, it can be done.
edit, P.S.: It's worth keeping a bare level of suspicions that there could be a common lower level cause, whether through inadvertent dependencies added or through hostile action.
Apparently the FAA outage was caused by the pressure to do things quickly, fueled by the AI urgency.
It feels like all the IT industry is feeling this pressure.
That's the preliminary post mortem from Microsoft.
I don't think we'll ever get anything honest out of the administration for the FAA outage unless Congress really wants to dig into it and make the results public.
Must be the water then
b) there is also a risk of sophistry and language games obscuring common sense truth
In the world where a state actor is engaging in active sabotage as part of a cyberwarfare effort, there is one shared cause.
Which exploits are responsible in each case may differ. Still one cause.
Do I think it's Iran?
No idea.
But the point is, none of us here have any idea at present.
- Massive Microsoft outage - FAA ground stop - Now this outage
Definitely a black swan at least? It’s not even the number of failures, it’s the scale.
AI is acting as a sudden precipitator for problems that have been latent and growing for at least 15-20 years.
In pursuit of maximally rapid innovation and reduced capital expenditure, the software industry adopted a "move fast and break things" mentality and homogenized code and service infrastructure over a few winning cloud vendors, SaaS providers, and code solutions/frameworks/packages/etc.
The pitch was that you could rely on downstream experts to do better work than you could, that the biggest vendors and publishers had the best experts, that have them invest the capital would look better on your accounts, and that their scale would give you the best value for your investment.
The caveat is that effectively all technology products have serious faults lurking in them, even with the finest expertise working on them (not that most actually had that), and that when everybody homogenizes over the same providers and solution the industry shifts from a bubbling froth of localized bugs to a heavily correlated house of cards where widely correlated failures manifest in widespread collapse and catastrophe.
Whoops.
AI is acting as a breeze against the house of cards, but lots of things could have and would have and did, especially as it continued to inflate year over year.
It's really hard for me to get a sense of how much outages are on the rise - versus every minor issue getting a lot of attention.
Ours was misconfigured as ourdomain.com instead of mailserver.ourdomain.com
Here's their announcement: https://www.spamhaus.org/resource-hub/deliverability/spamhau...
"on a long enough timeline the survival rate for everyone drops to zero"
I also don't see how you would ever end up in this situation if transacting via fb marketplace as you're not even allowed to have accounts that are not you.
My bank used to let me send checks via online banking (either electronic or mailed out physically by the bank depending on the recipient) until they adopted zelle and discontinued that service. My credit union still offers the check thing for free. That's an easy choice at least for me.
Banks only allow a couple $k of transfers per day anyway. The company behind Zelle, Early Warning, is a bank-owned centralized fraud detection network as the name suggests.
The ability to send wires from your account is riskier... I wired the down payment on my house in one go.
And your bank will always say it was authorized.
https://www.nytimes.com/2022/10/03/business/zelle-fraud-warr...
"Banks draw a distinction between what they consider scams — when a customer is tricked into sending money to a thief — and fraud, when an unauthorized person gains access to a consumer’s account and siphons off cash."
If you're drunk at a bar and someone unlocks your phone with your face and sends themselves your money, that's gonna be authorized from their standpoint.
Cash, check, wires, ACH, money orders are no different from Zelle in that respect.
You can’t drain your entire savings through Zelle in a single day (unless your entire savings is $3k).
not sure what you're trying to say. I can be stupid with cash AND in my digital life, the choice is mine.
You don't have to follow my advice, for sure.
The median account of money people have in savings is quite low; $3k is a life-changing loss for many. https://money.com/average-american-savings-account/
- Nathaniel Rothschild Nov 5, 2026
You also have full control of downtime and recovery rather the refreshing the MSFT status page.
People love to say this but can never point to who these experts are or what their expertise was.
They would/will never admit it is AI breaking everything either.
[1] https://www.mediaite.com/media/weird/flavor-flav-more-than-o...
The thing is, even if true, they'll never admit it. People who make these kinds of mistakes (if this is what it is) are also not very prone to honest bouts of introspection either.
The thing is - I also think AI can be used to build very high quality software. The problem, I think, is that the typical big tech/VC tech/MBA tech type company (who control most proprietary software at this point) have very little incentive to actually care about quality, and that was quite obvious even before AI.
I always defended this, but it appears maybe we don’t have enough developers that take pride in the work they ship. LLM coding made this pretty obvious.
So I speculate that a lot of problems with AI code are just that it is accepted uncritically because it builds and doesn't have any syntax errors.
Not really a great recipe for good code.
I don't, and I think your comment is rhetorically useless because you've been bullied into making this ridiculous concession.
Agentic coding could be a world-changing benefit, but the current way it's being sold and used as as a cost cutting measure -- produce software faster than ever before with fewer software engineers! Get rid of those pesky experts!
It is a C-level executive problem. With the FAA it is the agency directorship problem. Same issue.
These people, very literally, are trying to run the world with power point presentations and executive summaries drafted by sycophants. 70 years ago these people had actual experience doing actual work and trying to grow businesses and doing stuff at the ground level. Now they are given their positions through a combination of credentials, connections, and political maneuvering. None of which actually relates to how well they can or cannot run a organization.
When reality disagrees with what they get in their emails and presentations then it reality that is wrong. It is the messengers that are the problem, not their competency.
This is nothing new. But the failures in leadership compound over time, like interest to a unpaid debt. They start off with healthy growing companies and thanks to government protections and bailouts it takes decades to for these organizations to decay into total dysfunction. In the past these sorts of people would be removed and their assets reallocated to competent management through bankruptcy and competition. However that natural process is not allowed anymore.
All AI is doing is handing them more foot-guns.
They think they can do better/faster/stronger/cheaper with AI and if AI causes problems then it can just be fixed by paying for better AI. Which they assume is just right around the corner, always.
Just wait until they start to realize that it is
Don't want to get into a political debate here but I'll say that I think that open AI models and access to them is a benefit to society, while closed, proprietary models, controlled by tech billionaires are probably net bad for society.
Could you define 'proper training'? I've endured plenty of fucked-up shit from so-called 'professionals/ experts/ trained people' over the last 30 years or so.
I'm currently assisting a couple of vibe-coders improve their 'too easily misused...[and] widely distributed' (available on their sites, or github) tools - I mean slop - that benefit me in ways traditional companies have no interest in. And the updates are fast (daily, mostly) and accurate because, unlike jeremy with a pretty CS degree and a nice job, these people actually give a crap' and don't have a manager to say 'you can't do that'. Also part of being sane is accepting that the cat is out of the bag and big $oftware/ tool manufacturer better get a move on. Disclosure: my interests are not critical to anything; quite the opposite, but people are learning faster and faster with Ai learning faster and faster. Who needs azithromycin if a series 1's and 0's can accomplish so much more? For good or bad.
What a disaster.
No excuses.
[0] https://cointelegraph.com/news/doordash-stablecoin-payments-...
I don't take your assertion of certainty seriously. In my opinion, you are not qualified to evaluate the quality or security of code. You are looking at a series of related disasters and shouting that nothing is wrong. Please forgive me for not taking your perspective seriously.
Company process and how LMS are treated in the development process is actually of interest. Are they YOLO shipping code or are they reviewing it and using LLMs as a tool, instead of a replacement for an engineer? That is where LLMs bring danger.
I use Opus 5.5 and it can't generate code anywhere near as good as mine. I'm not an exceptional programmer either, I'm fairly middle of the road. LLMs remain bad at programming, despite the claims of AI bros that only within the last 6 months have models become good (which they curiously have claimed constantly for years).
Ugh. DownDetector's methodology is… not great.
Steam isn't down: https://downdetector.com/status/steam/
Amex isn't down: https://downdetector.com/status/americanexpress/
Visa isn't down: https://downdetector.com/status/visa/
AWS isn't down: https://downdetector.com/status/aws-amazon-web-services/
But each has a huge corresponding spike there.
"User reports show problems with Google"
https://downdetector.com/status/google/
As the Rick & Morty quote goes, "that just sounds like horrible methodology with extra steps".
Seems straightforward. They could add some qualifiers about the lower ranks somewhere, but that's the base value proposition and it works fine.
> you see the cause of the outage
What does DownDetector say the cause of today's apparent Amazon and Google outages are?
You can't source details that depend on technical knowledge from a non-technical audience, regardless of methodology. Downdetector is useful nonetheless.
You told me only "the top two, maybe up to four" slots count. Five are named in the headline. Only Chase has confirmed any kind of issue, hours later.
(In fact, the article has now been updated to say that Bank of America doesn't have an issue, and suggests it's a Chase-specific outage.)
Any time a major service goes down, all the others tick up because the people who use downdetector are absolute idiots.
The article writer then went to DownDetector and wrote a headline "Outages reported at Bank of America, Chase, U.S. Bank, Wells Fargo and Zelle" based on just that. As far as I can tell, there's zero genuine evidence for that.
Is there any other sign they are? That Google, AWS, Steam, Amex, Visa, etc. are all simultaneously having significant outages right now?
> Responding instantly to partial outages that never get reported on status pages seems to be one of DownDetector's biggest strength.
I've predicted all fifty of the last one recessions.
DownDetector considers someone looking up "is Visa down?" a report, even if it isn't.
Click my links and it contributes to those "reports".
One provider has a major issue, and people go there and click into non-down companies, which is seen as a spike.
On the other side, does a user google "is Visa down?" going to a website and clicking "yes I think visa is down" any better?
Normal people Google "is Chase down?", find DownDetector, and click it. (Note that the article has a Chase rep quoted saying they do indeed have an outage, and none from other companies.)
Some of them then click into the DownDetector home page and browse around. Suddenly everything is spiking.
Every time I've seen any one major company with issues, DownDetector has spiked across the board. As it is now. Google and Amazon are apparently down.
Really? It seems to me you have to click a specific aspect of the service to report as down once you land on that page to report an outage.
> We process tens of millions of problem reports each month and cover over 36,000 services in 72 countries.
Ask a few normies if they know what it is, let alone have filed a report.
https://downdetector.com/about-us/
> Downdetector collects reports from users and monitors other online sources for service issues.
Or look at the comments on any of the pages. "My card declined"; "18yr femboy down for DM".
Steam/AWS, I got nothing.