Well, this was way overdue. Sloppy, crappy systems exposed to the public internet, no proper security, thoughtless software implementations based more on “how fast can we deliver?” instead on “is this product actually good and up to the task?”.
We just need to stop all online advertising. It seems most effort in the last 20 years was to find a way to infiltrate every corner of your computing experience by injecting an advert in there somehow. There are lots of secret, hidden pathways for adverts to follow that AI is now capitalizing on. /s
None of the recent outages was caused by a hack as far as we know. The hundreds of websites using Optimizely also didn’t get hacked. Only Optimizely got hacked. So we have one confirmed hack. What “sweeping cybersecurity attacks” are they talking about?
There are more critical CVEs being released every day at an insane rate, and there are automated AI tools being utilized to exploit them at an equally insane rate. Small/medium orgs are experiencing breaches but it doesn’t make the news, and if they’re on top of all the CVEs and are actually able to patch or implement mitigation steps, they are spending significantly more time on opsec than they’ve ever had to.
Not everything makes the news, and orgs with competent technology teams contain breaches quickly and restore from clean backups and move on without even announcing it. But the cyber security attacks are happening at higher rate than previous decades.
If you are writing an article about it would be great to back claims like that with some facts.
Nioce
Honestly this is not a bad time for all this to be happening. Better now then while actually in a war of attrition. I don’t believe Russia has any part in revealing their enemies weakness’ right now when those exploits would be far more effective in the very near future. It’s like a grey hat campaign it feels like.
The hugging face incident was caused by credentials being found in github by the AI agents. The freezer attack was because they never changed the default login from admin/admin to something else. The grounding of all flights was due to a misconfigured router in Arkansas that started blocking certain packets. AI has found RCEs and other serious exploits but they are all under the control of the frontier models for the time being while they get patched. This is all to say that fear is being sold and sort of manufactured to create a grand narrative where no such thing exists. There are many reasons for this, control and power is usually the root cause. In this case getting control over AI development is defined in National Security terms, but it’s really just more about who is going to control whatever power comes from this technology in the future. And it’s not often you see the kind of triple tap, 1-2-3 coordinated release to take out Jensen Huang and sort of get control over the central bank of AI. And so this over hyped media narrative is about shoring up public support for Government regulation and control, along with a sort of media/PR campaign where they understand that fear sells the capabilities of these frontier models to investors for all of the upcoming IPOs.
Eventually though, once open source, smaller and more contained models targeted for cyber security, or when nation state level actors start putting resources into weaponizing the scanning and usage of exploits, then we will see a quick ramp up of a cybersecurity arms race. If you kind of game theory it out, all software will have to be constantly scanned by the most modern advanced models to find any exploits or bugs the prior lesser advanced models missed. This means that most software in the not too distant future will have to be scanned and maintained by frontier model companies, because any human written code might be too insecure to be able to be deployed. In fact a sort of regulatory moat for the frontier labs could be that code deployed in the future through app stores will have to pass an AI scanning certification process by one of the big model companies. But if all of these ongoing regional wars keep escalating and the U.S. gets drawn into them more, and some of the countries on the other side start wanting to use weaponized exploits their models find but the U.S. models miss, then we will be in a situation where actually selling this fear will be a reasonable response, instead of whatever it is right now.
Thanks for reading my wall of text
And this shit is why all important things need functional analog alternatives.
I have already countered this move by having a website absolutely no one visits. Checkmate losers.
I decided to go back to college to get into cybersecurity. I changed my mind, fuck it.
I’ll just get a bag of marshmallows and sit back and watch.
It’s scary how a single service failure can affect so many websites at once.
Are we really though? Or is it just crap code.
There is a lot of code with vulnerabilities from decades of coders not doing a bad job, just not doing a perfect job. Regrettably now we have weaponized vulnerability exploiters at the hands of everyone on the planet.
I think this is it, years of underinvestment, multiple rounds of layoffs, shipping the absolute minimum viable product, all in the search for maximum profits. Then shocked Pikachu face
Mostly just crap code and/or shitty corporate IT management.
A bit of both. And a dash of general cheapness, too.
AI, for all the bad, is able to quickly exploit flaws. Mostly already known ones. Admins are not always keeping things up to date, when there’s even a admin. Infrastructure is badly done, having gaping open doors that only asks for someone to enter. Competent people all around, able to detect, act, and fix, are removed because who needs this.
AI, like all computer programs, is really good at being persistent.
From the beginning of computer programming, the ability to engage in loops to exhaust every possibility was always going to be faster and more comprehensive than what humans could, or were willing to do. You could naively try every combination for cracking PINs. Then came more sophisticated loops that would try every combination by first going through known password patterns (common dictionary words, common numerical combinations, etc.). There were rainbow tables computed to where certain computationally expensive processes could be pre-computed before a leak so that a leak of hashed passwords could be more quickly exploited.
And there are tools scanning the web for all sorts of known vulnerabilities, being much more comprehensive and systematic. If you have a vulnerable machine exposed to the internet, an automated hacker might compromise it.
The current state of AI-assisted hacking includes tools for analyzing patches and CVEs to reverse engineer exploits for unpatched machines. So everyone is now in a hurry to apply patches as soon as they come out, lest they be left behind as hackers adapt their scripts to keep crawling for recently patched vulnerabilities.
Plus there are more tools for searching for vulnerabilities, either in code or at some sort of other attack surface. The number of vulnerabilities being patched right now is pretty high, which puts just higher pressure on all of the processes to more aggressively develop quick patches for known vulnerabilities, install patches as soon as possible after release, etc.
Throw in an ecosystem where programmers might have been laid off, or teams are understaffed, or using AI tools themselves and have less of a firm grasp on their own codebase, and that’s just a recipe for less reliability.
deleted by creator
AI, for all the bad, is able to quickly exploit flaws. Mostly already known ones.
Absolutely agree. It’s basically a fuzzer with a big database of exploits. Of course it can find exploits.
It’s one area it makes sense to me, but I’m pretty sure you could do even better with a specifically trained model rather than a general LLM.
Regardless, you still need buggy code to exploit.
you still need buggy code to exploit
Hasn’t that always been a near-unlimited resource though?
deleted by creator
AI has led it to a previously unthinkable level of buggy code.
Where is Barron Trump when we need him?? He can turn a computer on and off like no other it’s incredible!!
deleted by creator
Some jokers have named this phenomeon “The Crapilarity”. I am not sure whether it is written with “h”, but I guess there will be a few tasty roasted chickens on the road.
(And probably more work for good software engineers than anyone was dreaming of…)
“Te Crapilarity” sounds weird, so I am pretty sure the “h” is there! :)
Oh, to be precise, I was wondering whether it should be written with two “H”: “THe CrapHilarity”.
I think they meant craphilarity - crap-hilarity vs crap-singularity
Martial law in 3, 2,
Incorrect.
Martian law.
It’s funny, looking at each of these things on Hacker News, every time it’s incompetence to blame. So that’s what the tech industry is saying to each other. This receives a lot of “well, yeah, you run A B tests and load 10mb of JS on every page, this is an obvious risk associated with that.”
Edit: yes, their bias is towards “yeah, don’t replace me with AI. Let me do my job better with AI.”
Including, that if things like Outlook being down for a day were cyber attacks, how fucking clownshoes are they? Outlook down for a day? Apple down for a few hours? Not exactly some crippling “Cyber 9/11” we’ve been told to fear for 10 years.
Risk is profitable. Any time you calculate profitable risk over odds of something taking down a system, the c-suite will choose profit.
It’s funny, looking at each of these things on Hacker News, every time it’s incompetence to blame.
Yep, people are still relying on CDNs even though they provide no real benefits, and then not even enabling the security features you’re supposed to use with CDNs.
Subresource Integrity, the ability for a site to use a CDN while blocking malicious versions of scripts, which would have blocked this whole attack, has been supported by browsers for 11 years at this point.
https://developer.mozilla.org/en-US/docs/Web/Security/Defenses/Subresource_Integrity










