105 comments

[ 5.1 ms ] story [ 52.1 ms ] thread
"Microsoft attributed the burgeoning patch counts to vulnerability discoveries aided by artificial intelligence."

If only real intelligence found the fucking things instead.

As ye sew, so shall ye reap!

An employee just got phished by adding a number to a legitimate deviceAdd login route that bypasses 2FA and adds a device with full access to office and mail

Probably working as intended...

Sounds like one of ADOs recent security misconfiguration vulnerability announcements. The customer is blamed, for not quite hardening everything the right way, when ADO config is... A sizeable task.
I wonder how many bugs will be introduced with these fixes...
No bugs, only intentional backdoors
Fix one and add two.

There used to be a brilliant weather app here in Oz back in the early days of iOS. I always loved the update notes the author provided. One was "Fixed one grammatical error and introduced another one, can you find it?"

It depends on how good their testing practices and code review / auditing are. I want to believe they are some of the best in the industry, but that's making assumptions.

But jumping to assumptions that fixes introduce bugs is a bit rash and assumes incompetence / unprofessionality / unmonitored AI usage / kneejerk bugfix processes.

Microsoft (in)famously fired all their QA people like a decade ago.
Most Saas delegated QA to their clients...
I remember some research from IBM that showed that every two bug fixes introduced another bug. Effectively your queue for bugs is twice as long as you think.
Sounds like a lot but compare it to Edge also being patched for 428 Chromium CVEs this month.

If 20 years ago you told me a single piece of software had 428 vulnerabilities I wouldn't have believed it.

If Chromium has that many security bugs, perhaps the move fast and break things approach of spraying diarrhea masquerading as code into a keyboard needs to be reexamined.

20 years ago software wasn't as much battle tested as today, had way less feature set, was less connected to the internet, and etc. 428 CVEs looks small, assuming not all have CVSS 9.8 or something.
It was more tested as real testers were testing it. Nowadays, AI just checks the code.
Is that true or an assumption? I don't think Chrome and co would discard their decades worth of automated testing practices in favor of AI.

Actually I don't need to think / assume, it's an open source project.

Chrome? That is a younger project and it was much less automated back in the day.
20 years ago was 2006, I can assure you we were very consciously aware of things being permanently connected to the internet by then.

That said, sure, it had a fraction of the features back then, and only a fraction of the world population was connected to the internet.

>That said, sure, it had a fraction of the features back then, and only a fraction of the world population was connected to the internet.

And the world was better for it. Connectivity and internet are not inherent goods. They can be used for good purposes but it is hard to argue that has been the mean.

>features no one asked for

Google asked for them. That's all that matters.

20 years ago a malformed packet to winsock would crash the computer, 5 years later installing win2k on my buddies computer (no router/firewall) a few minutes after we finished the install "windows will reboot in nn seconds" whelp time to re-install without a network connection... we've added a lot of layers since win2k, mostly in the name of ease of development, and I don't feel like we've met that goal but we sure found a way to get a million monkies behind a million typwriters, and now we're aiming to replace the monkies with simulated monkies. Time to smell my fingers and fall out of the tree ;-D
Chromium is probably 30 million+ lines of code, and we generally see that as things get more complex its even easier to accidentally write code which can be exploited.

If it has 1 vulnerability in every 10k loc of code we'd be talking about 3,000 vulns (with no churn) - we used to care about defect density, and most software wouldn't go more than a few hundred lines without SOME bug, whether that's a "vulnerability" is often a layered question.

Chromium is a three-decades-long project (origins coming from Webkit, which comes from KHTML, which was 1998) - I can assure you that does not meet the "move fast and break things" phrase you're trying to smear it with.
It would be nice if microsoft had windows update for .net, visual c++, office, windows, edge ... just all their software in one updater, but that would be too easy...
It did work that way for .NET versions but the patches and upgrades caused too many bugs and incompatibility. Folks would install old .net versions anyway.

The pattern moved to packaging in all your dependencies.

Winget/Microsoft Store etc could auto-update your apps even with packaged .NET DLLs, though.

It's (somewhat) possible with winget already.
IIRC winget is another mechanism with a CLI front-end that resembles a package manager. What it does is either hook into the windows store (another channel) for packages, or points to an installer which likely runs silently in the background, for example 'winget show adobe.acrobat.reader.64-bit' or 'microsoft.vcredist.2015+.x64'
It seems like bug hunting might be the one area where AI is actually making the world a better place.
How many were introduced by misuse of AI coding/vibe coding though?
How many were known, and put on the roadmap because war got hot?
highly unlikely for many of them. SharePoint, bitlocker, Active directory, hyper-v, rdp, DHCP and MSMQ are all software/technologies that have decades of history and long pre-dated LLMs. seriously, do people not realise it was entirely possible to write insecure or bad code before LLMs?
Especially if they made heavy use of offshoring, which I would bet they did.
> seriously, do people not realise it was entirely possible to write insecure or bad code before LLMs?

Some of these threads make me think every line of code written pre-LLMs was apparently perfect in all ways. Feels like romanticizing the past.

It’s like people don’t remember the whole outsourcing trend and all the awful code that came from that.
Don't forget the damned interns!
Hey, I wrote slop at Microsoft way before it was cool.
It's funny because SharePoint and AD are so god-forsakenly awful you would think they're vibe coded if you didn't know any better
It is hard to tell, the code may genuinely be decent quality or not.

That is the issue with vibe coding. Increased output but reduced understanding. So if something does go wrong, one has to hope that there is still enough understanding to address it quickly.

For what it's worth, at my workplace AI has uncovered quite a few issues that have been there for a decade or two and survived countless rounds of careful reviews, external security analysis, pen testing and so forth for all those years.
The distinction I'd draw is between AI-assisted and AI-generated. Using AI to write isolated functions you understand and review is different from prompting your way to a complete system you can't debug. The second case is where you get surprising failures at runtime that no amount of linting catches.
Vibe-coded apps probably have loads, but mostly because they're using less capable models than the people who're doing the bug-hunting. Once vibe-coders are using models like Mythos too you should expect the number of bugs in vibe-coded apps to collapse quickly, because the LLM will write the bugs but will also fix them (assuming the system prompt tells it to.)
For some reasons, models are blind to their own output...
One reason seems obvious/intuitive: because their own output matches their own biases, that is, is a direct result of their model walks.
I think this could be true if you use a single Claude context, since it has its own reasoning in the context.

But a separate code review agent does much better, in my experience.

Not my experience with my homie Claudius.

The code review agent usually has feedback to be resolved before committing, which includes bugs and unhandled edge cases. Sometimes the primary context is understandably embarrassed.

Sometimes it truly be your own people.

Is it worse than what humans were doing on their own anyway?
Well yes if they expect to find and correct more bugs every update which is pretty much what they are saying.
If AI ever manages to make something half as atrocious as Windows XP that would probably be the first proof of AGI.
99.9% of people complaining about AI making the world a worse place would be fully happy with AI if they shared in the economic benefits of automation.
Not if it ends up deskilling society and taking away what brings us meaning in life.
(comment deleted)
I don't see why AI would deskill what you love to do. People still do embroidery even though mass manufacturing exists. If you love something you would continue doing it irrespective of automation.
Yes. I love eating pieces of string. I'm partial to wool myself, really filling.

I suppose you're a cake enjoyer, miss Marie Antoinette?

For a lot of people it’s not the job that is rewarding it the role having a job gives them in life and home life. Financially contributing to the household through earning it through work is a meaningful and rewarding thing that can define the near total of how good you feel about yourself thing even if you don’t like the job?
(comment deleted)
But do people make a living doing embroidery by hand? Or is it more of a hobby?
For better or worse this round of "automation" will hit harder than most others because it comes for what makes you you. Your brain, not your body. The industrial revolution replaced your body, this one could in theory replace you entirely. And it isn't coming for some job, it comes for most. It's not just a hobby but being able to do the job. You are replacing 10000 types of jobs with 1 type: AI prompter. We'll probably have an AI just to help ask for the right thing.

You have transferred all the skills and knowledge to the AI. When the skills are gone, who's going to teach you to do any non-trivial job? What will give you any sense of accomplishment when all you do is ask the AI but have no capability yourself?

Even if you have a guaranteed income, because "your AI" is getting paid for the work, where will you be with close to 0 contribution to anything? Maybe we're overreacting and this will never be an issue. I know we shouldn't take cues from fiction to predict reality but it's hard not to picture a world with a combination of Idiocracy and Wall-e (or the famous "Paradise" Matrix) where we decay because the change is so fundamental that we aren't ready to adapt.

Yeah this round replaces both body and knowledge.

Basically what is left is internal politics and cross company dynamics until we get to the point where a company is self autonomous.

(comment deleted)
I've been benefitting from automation my entire life. I don't see why anything would change when that automation is driven by language models.
for your first couple of B you stop complaining about anything
The bonuses SK Hynix staff are getting might vindicate that point of view. Of course, that's from AI investment, not revenue itself. They are the extremely successful shovel manufacturers, who for once have managed to leverage their power collectively as employees to benefit from a one-off like this.

Perhaps it has always been the case that people would be happy if they shared the benefits, but that is not how the world run by billionaires works.

"People complaining slavery makes the world a worse place would be fully happy if they were slave owners"

Probably, but just proves people can ignore bad things when they benefit them.

Not what it pretends to prove, that the thing isn't bad.

(comment deleted)
It was like a bunch of mythos scans through and through which then generated the reports for everyone to implement, not sure if in all orgs though. Mythos was great as it came from the top, i.e. a clear incentive. I think bug reporting otherwise does not reach the engineers unless an incident is raised by the customer care against a responsible team.
Bad guys can use AI too. Not sure about the net result.
Depends. If the AI fixes don't introduce new bugs or unnecessary complexity (where bugs like to hide), then yes.
There's also the weird scenario of a reporting bias where instead of admitting to a bunch of vulnerabilities, you can frame it as "look at how useful AI is".

We even have companies implementing solutions for ffmpeg vulnerabilities themselves instead of just handing them the vulns to fix themselves.

It's very possible the tide reverses if it's not in people's interest to advocate for AI anymore, so we better not get too used to it just in case.

It's nice that we currently have an alignment of AI advocacy and infosec, though. Maybe Microsoft can even point their AI to their questionable UX and UI practices next.

I am curious, they say we should expect a trend of more security patches every update.

It this is true and Microsoft devs are using agents it means that AI is doing a shitty job and is introducing more bugs/vulns per month than it fixes them. Otherwise you would have had a lot of bugs/vulns fixed in the first 2 security updates then a steady and significant reduction every month because any new code would have been scanned and fixed before release.

They should patch that Global Device ID thing

:^)

This is great. Now ask Mythos to make windows suck less and let it go crazy.
Deletes Windows 11

Installs Windows 7 with new patches

What would be the point of any of this if the registry is still there? It'd just a particularly shit version of linux window management that happens to run games well
The registry was never an issue tbh, it's just a database that some companies decided to fearmonger about to sell you "registry cleaners".

But if I'm wrong, please do point us in the direction of known issues with the registry.

The registry absolutely is an issue—it's everything bad people make the command line out to be except ten times worse because they haven't bothered to fix it in forty years. You need to cooy and paste an incomprehensible string off the internet just to set the functionality of the caps lock key or rewire a file extension. Just straight up dogshit softwareif you weren't raised in the cult
Calling the registry a database makes it seem better engineered than it really is.

Running queries on the registry like SQLite? Nope. The Windows registry internally from what I've been able to read is a weird reimplementation of a file system (a trivial database I guess but a real database lets you index on things other than name). It's compact but as far as I know completely unofficially documented.

(Most data structures with unique names or ids for each data item could trivially be considered databases, but generally if you say something is a database, you typically expect to create queries more sophisticated than "select data where name is X" and also expect to be able to create or use separate indexes to support those queries.)

The reputation of brittleness and danger of the registry is because Windows stores a lot of configuration parameters there that the kernel uses as boot, but because it's not a plain text file, comments explaining what the settings do can't be right there in the same place you're editing.

Installs Windows 3.11 with no new patches
It would certainly make sense to sick it on all those crash reports and complaints. All the obscure 0x error codes that end up with you being punted to the forums for 'community support' and being told to reinstall everything.
Those errors in 99.9% of the time are hardware related or driver related. I've personally not seen bsod or had a windows crash in years. Last week my on board Lan card started erroring out but windows didn't crash, I just lost internet. Switched to wifi and moved on.
How many are chained, and how many patches are defense-in-depth after discovering chained paths to that flaw?
Title is not correct. Microsoft didn't patch a lot of this, they're reporting patches for dependencies that other people patched and Microsoft are inheriting.

For example, Mariner (now branded Azure Linux) is a Microsoft-supported Linux distribution. So in this list of 570 vulnerabilities, Microsoft have reported 100 vulnerabilities inherited from all sorts of open source software projects included in their Azure Linux distribution. The OpenSSH vulnerabilities are described in better detail at https://www.openssh.org/releasenotes.html where it implies 2 vulnerabilities were detected with Swival Security Scanner (using LLMs) and another 6 by other researchers/companies (using undisclosed methods). Swival have published the output of their AI-detected CVE-2026-59996 (scp(1): when copying files between two remote destinations, do not allow a malicious server to write files to the parent directory of the intended target directory.) at https://github.com/Swival/security-audits/blob/main/openssh/...

If you want an easy way to view these I made https://wofa.dev to keep track of windows updates and security patches in a single place
It is time to abandon Microsoft.
Microsoft has its place but I’m not sure that includes personal computers anymore.
While I do think it's wise for people to start looking at alternatives, I don't think this specific news should sway one way or another.

Windows (like Linux or macos) contains an enormous amount of code, and with large code-bases you're certain to have security issues.

Finding these security issues and fixing them seems like a good idea, no matter how much you love or hate Windows.

Code can exist without vulnerabilities. Verifiable proof exists. It is hard, but it is possible.
If I find a bug in very important MS products, I usually do report them using their shitty feedback app. Not once has any of that yielded anything.

Maybe a way to find tons of high impact bugs would be to let MS developers access those bug reports?

There's more direct channels to report security vulnerabilities - which this about.
Perhaps the feedback boxes they put in every app should actually be read. That’s a them problem.
Those boxes are not there to generate feedback. They are there to placate the user by implying that their voice will be heard.
I reported 3 errors/feature requests in Oura Ring app. After going through the reporting process I felt much less irritated at the bugs and smiled to myself thinking that must have been the purpose of the reports.

To my astonishment, 3-4 months later they fixed the errors and rolled out the feature I asked about.

WTH, people read those reports?

I know people do, I often get in touch with support for various services, knowing that in the sea of people not giving feedback, your comments mean a lot. But up until recently both the Oura App and their support was very mediocre.

Where's my enshittification?

Gets me wondering if it is real or they just got a backlog quick fixes dropped so it looks like AI bug hunting really works to prop up their AI bullshit narrative.
Tha attack surface of Windows is infinite

As one door closes, another opens

"Features"

"Fixes"

Job security

Security holes

The threat model is Microsoft