CVE-2026-99312: KVM Guest-to-Host Escape
127 points by Kozmik1 3 hours ago | 76 comments

jacobgold 2 hours ago
Our infrastructure will become so much more secure but we're in for a bumpy ride through 2027.
reply
dvh 2 hours ago
Not necessarily. Only if you assume that no new insecure code is added. If ai found 1 security bugs in old code but ease of code generation with ai adds 2 new security issues (in unrelated code), then we are getting less secure.
reply
tyingq 2 hours ago
Is it a given that it's as good at outputting secure code as it finding insecure code? That leaves a fair amount of room for people not baking in deep review steps. Also mentioned elsewhere, but if velocity is way up...that ramps up bugs anyway.
reply
MajesticHobo2 2 hours ago
No, it's not a given. AI always finds security issues in the things I just had it write.
reply
hparadiz 57 minutes ago
This is true for human written code too. This is why in the future all written code MUST be run through an internal security sweep before being given to your harness for writing. That's just basic due diligence at this point.
reply
vbezhenar 19 minutes ago
As long as model providers deny access to security-capable models, it can't be basic diligence.
reply
hparadiz 15 minutes ago
This is something agents already do as part of their normal work. Providing code that you know is vulnerable is in fact a dereliction of duty and negligence. Don't provide a POC. Simply that as it produces code a background agent just checks it without showing the work and points out potential problems.
reply
catlifeonmars 36 minutes ago
Yes but LLMs famously produce vastly more code than humans. My hot take: LLMs will produce more insecure code than they can catch, and will also find ways to smuggle insecure code to avoid extra work.
reply
SOLAR_FIELDS 2 hours ago
Do we have any evidence around the rate that AI emits security bugs vs humans previously? You would think that perhaps AI would be better about this, but perhaps not.
reply
dvh 2 hours ago
I'm making no claims. I'm just saying that contributing before ai was difficult so people rarely did it. Now that contributing is easy, more code will be added. More code = more bugs. Always.
reply
voakbasda 2 hours ago
Yet the opposite seems to be true. Many projects now shun new contributors due to AI.
reply
punchmesan 49 minutes ago
Eh I'm not sure that's an indicator that the opposite is true. Github's vastly increased commit activity is the best indicator we have of this. https://www.tomshardware.com/software/github-report-highligh...

Sure, some projects are trying to be responsible or are trying to keep waves of slop PR's from drowning them, but that is by no means an indication of a general distaste for AI code. It's been largely embraced.

reply
hparadiz 56 minutes ago
I've run Daybreak against my human written pre-AI code and found so much stuff that I don't trust human or AI written code. It all MUST be audited by a security sweep AI.
reply
the8472 2 hours ago
Even if Fable and Astra are superhuman at identifying and fixing bugs that does nothing when projects get bugslopped faster and cheaper by some low-end model.
reply
tptacek 2 hours ago
I hate these meta conversations on crunchy technical topics and feel like we should be digging into the KVM security model more than this general security prognostication stuff. Having said that, I will hypocritically further this debate by pointing out that the determinant of software security moving forward is not so whether AI is drastically better than humans at avoiding common vulnerability patterns, (or instead just a little better). Rather, it's that AI will make it much easier to move subsystems or whole projects to memory-safe programming languages and towards formal verification.
reply
j45 2 hours ago
Existing code will have security holes that haven't been found yet.
reply
InsidiousWombat 26 minutes ago
...if the pace of both fixing and finding bugs got accelerated, all we're in for is machines being on average more vulnerable (because patching takes time and effort and more bugs increases that)
reply
TZubiri 33 minutes ago
Have you ever been told that you are problematically optimistic?
reply
gerdesj 2 hours ago
404
reply
veverkap 2 hours ago
Same for me - did they remove it?
reply
fwsgonzo 2 hours ago
Don't disable nested, it's still better security than everything else, and having it on means access to real hardware virtualization. Second, the article is AI-written. It really should be rewritten by a human.
reply
xmodem 2 hours ago
Outside of development and CI, I have yet to hear a compelling use case for nested virtualization.

Agree that the article is downright painful to read though.

reply
JoshTriplett 2 hours ago
> I have yet to hear a compelling use case for nested virtualization.

AWS instances are slow to launch; they take whole seconds. Nested virtualization allows running something like firecracker/crosvm/rust-vmm atop a normal AWS instance, without needing a bare-metal one.

reply
tptacek 2 hours ago
If you're segregating internal workloads, by all means use nested virtualization on EC2 or DO or something. Your threat model is different and you probably have a lot more room to maneuver when a KVM escape is announced, because you have some level of trust in all your workloads.

If you're running multitenant for mutually-untrusting customers, that excuse goes out the window.

reply
JoshTriplett 2 hours ago
One distinction I'd make as well is "are you running more than one customer's workload on the same system" (yikes, don't), vs "are you using nested virt as additional protection rather than running the customer workload directly on an instance, but one instance only gets jobs from one customer".
reply
tptacek 2 hours ago
Another good point, but this one is maybe a little less dispositive than it looks like: it's very difficult to engineer an orchestration system that reliably segregates worker physicals from each other, so I'd rate the likelihood of a game-over situation after a guest->host escape as very high regardless of how yuo schedule workloads. It's sort of the same situation as how SSRF's are usually game-overs (you're pivoting from a hardened attack surface to a much softer one) but even harder to engineer again.
reply
JoshTriplett 2 hours ago
That's fair. You'd want a system designed around the notion that every customer instance must be isolated from each other without the nested virtualization.

(Aside: my current guess at the appropriate security posture towards Nitro Sandboxes, in the current era of vulnerabilities, is "decent, but probably has security holes too, and is less widely scrutinized, don't count on it either". Thoughts?)

Any recommendations for good guides on that, besides "design very careful IAM policies and security groups, harden any common services they have access to, and ideally run them under separate AWS accounts"?

reply
necovek 34 minutes ago
Does this hold for the likes of Amazon?

Do your lambda functions always get a per-customer dedicated instance or is firecracker distributing them so that multiple customers can have VMs on the same actual box?

Sounds like it would be pretty expensive for Amazon to dedicate an instance to every customer, especially for those with small workloads.

If that's the case, why would it not hold for Amazon but would hold for everyone else?

reply
swiftcoder 28 minutes ago
> Do your lambda functions always get a per-customer dedicated instance

At some point in the past they did, for pretty much exactly this reason. I'd assume they don't anymore (I last worked there a decade ago).

reply
TZubiri 32 minutes ago
? If you don't use it, disable it. Simple as that
reply
rootnod3 2 hours ago
> It really should be rewritten by a human.

That does not seem to be what HN has been thinking for months.

reply
praseodym 2 hours ago
The blog post is very difficult to read, I suspect that unfortunately it’s fully written by an LLM.
reply
xlayn 2 hours ago
doesn't change the value of what's writen, if true and actionable that's all that should be needed
reply
manbash 36 minutes ago
It has "potential", not "value".

Also in this case, a "missed potential" if it eventually fails to reach anyone.

reply
sylos 48 minutes ago
Value is more than just iron something is made from. How it's shaped and presented is just as important
reply
tptacek 32 minutes ago
No, it is not "just as important" as the underlying information.
reply
tsunagatta 18 minutes ago
It actually might be far more important. You are a lot more important than the molecules that you are made of
reply
drum55 2 hours ago
If it was valuable a human would have been able to communicate it in a few lines of text. Using a LLM to blast slop devalues everything and communicates little. A drunken twitter rant has more ability to convey information than sloppy blog posts with 3 pieces of information per sentence and a load bearing attitude.
reply
Ylpertnodi 30 minutes ago
Get hip to it, daddy-o.
reply
cluckindan 59 minutes ago
Just ask the clanker to read it buddy, you’re already behind /s
reply
bigfishrunning 57 minutes ago
It's inverse compression!
reply
catlifeonmars 39 minutes ago
It kind of does, the OP could have just saved themselves the tokens and posted a 130 word summary on x/twitter that would have been just as useful.
reply
InsidiousWombat 27 minutes ago
well it's 404 now
reply
Betelbuddy 2 hours ago
The LLMs find the security issues, the LLMs will write the reports and the LLMs will fix and patch them.

I guess its time for all of us to step aside, leave space for somebody who knowns what they are doing, and go fishing...

reply
cduzz 6 minutes ago
We're all living in the story of mel.

At first, there were just big logic blocks and storage that did as they were told.

Later, in hardware, there were all sorts of optimizations built in where the logic would do its own thing -- so you'd get caches and speculative execution and even on-the-fly conversion from one instruction set to another.

This abstraction was also piled onto the runtime -- you'd write machine code, then assembly, then higher level languages; eventually you'd run in interpreted or just-in-time generated bytecode (hello to Java from UCSD p-System from the 70s and Burroughs machines in the 60s). And of course... operating systems...

Now, we've got machines reading source code (and machine code, and everything in between) and they've got some ability to thing the thing to make it thing. Sometimes correctly (enough to pass existing tests) and sometimes without the ability to express to the human what the thing they thinged was.

But, we're all living in the land of mel and for some it's always been better to go fishing than to pretend to participate.

reply
praseodym 2 hours ago
And this way only LLMs can learn from these vulnerabilities, making it even more likely that the next one will be found by an LLM.
reply
catlifeonmars 37 minutes ago
LLMs currently don’t learn, at least in the way most people mean, but I assume you mean the trainers will incorporate the knowledge in future releases.
reply
JeremyNT 9 minutes ago
And now it seems to have been pulled (getting a 404).

Perhaps they are rewriting it with less slop?

The CVE itself: https://www.cve.org/CVERecord?id=CVE-2026-99312

reply
biotechbio 2 hours ago
Glad to see this is the top comment. This is written by Claude, I know because I talk to Claude for 8+ hours a day, and it says stuff like this. I stopped reading the article because of it.
reply
arde 2 hours ago
That's a 404 now.
reply
cloudie78 17 minutes ago
Nested virtualisation.

Just don’t.

reply
gbrindisi 3 hours ago
Thinking about agents sandboxes... it truly seems like the time is ripe for unikernels
reply
wmf 39 minutes ago
Unikernels that run on KVM aren't going to protect against KVM escapes.
reply
rvz 15 minutes ago
Maybe the problem...is Linux itself.
reply
adverbly 2 hours ago
Well this sucks. So this needs a kernel fix then? Looks like I can disable nested passthrough manually, but am I stuck waiting for downsteam fixes at this point then?

Honestly, this gives me pause on the whole "VM is sufficient for any agentic sandbox environment" theory.

reply
tptacek 2 hours ago
Another nested KVM bug. We debated enabling nested KVM for many years, and held the line every time, because the only post-2020 KVM vulnerability we knew of required nested VMs. That decision aged well: we've gotten at least 2 more nested KVM escapes in the last year.

Don't allow nested KVM.

reply
rwmj 2 hours ago
Certain Windows features need it unfortunately. Virtualization-Based Security (VBS), Credential Guard, Hyper-V (obviously), and I'm going to say WSL2(?)
reply
tptacek 2 hours ago
Right, sorry, that's a fair point (and Josh Triplett made another fair point earlier). I'm sort of implicitly locked in to the "hosting company" use case here: services running mutually-untrusting isolated workloads for customers.
reply
Ungh 2 hours ago
reply
tptacek 31 minutes ago
Theo was wildly wrong about this, but he wrote it pithily so it's never going to die.

https://news.ycombinator.com/item?id=48883089

reply
Ungh 19 minutes ago
Well he's not really. It's just useful and shit!
reply
kristjank 35 minutes ago
There is a Theo quote for everything, isn't there...
reply
bigbuppo 41 minutes ago
Looking forward to Ubuntu finally rolling out an updated kernel for this in the year 2057.
reply
LelouBil 13 minutes ago
At this point, I'm just hoping the attackers get choice paralysis when thinking about which exploit to use.
reply
ArcHound 2 hours ago
Surely, we'll run out of critical CVEs in open-source someday. Seems it's not today.

In the meantime the patching will continue until security improves!

reply
j45 2 hours ago
Doubt it, but there will be a big spike for a while and then slow down.
reply
nubinetwork 2 hours ago
I guess its time to shut off the internet.
reply
calvinmorrison 2 hours ago
what is the mitigation.
reply
stilldavid 2 hours ago
Look in the article under the headline "Mitigation".
reply
pixl97 57 minutes ago
Page has gone 404.
reply
j45 2 hours ago
The cloud is somebody else's computer.

Mitigations are layered, first not using a shared VPS, and instead running your own bare metal server with something like proxmox to run the containerization.

Those containers could be susceptible to issues between them too, but you at least control the environment and can put more checks and balances in place.

Shared cloud and vps is going to become bumpy and more insecure the next little while. There's no way around it.

reply
k0morebi 2 hours ago
404 now
reply
pixl97 57 minutes ago
Confirmed, wonder what happened here.
reply
FLeXMurphy 35 minutes ago
Page is now 404'd.
reply