Solved The third major Linux kernel flaw in two weeks has been found - thanks to AI

Borg 386 · May 15, 2026

And yet again.....

Another bad Linux kernel bug has appeared.

Fragnesia can give unauthorized users root powers.

More open-source security bugs are likely coming.

According to Linus's law, "Given enough eyeballs, all bugs are shallow," is fundamental to open source.

Unfortunately, thanks to AI bug-finding tools, such as Claude Mythos and OpenAI Daybreak, behind most of those eyeballs are AI engines, and they're proving to be much faster at finding security problems than human ones.

The third major Linux kernel flaw in two weeks has been found - thanks to AI

AI is exposing Linux security holes faster than developers can patch them. Fragnesia is the latest. Here's what we know about it.

www.zdnet.com

andrew129260 · May 15, 2026

Yes, despite what some people say, AI is here to stay, and this will keep happening.

It's Internet 2.0.

Ai will bring lots of good, and lots of bad.

Finding these flaws are good imho, as long as they are patched later. The bad news is cyber warfare will rise highly.

Borg 386 · May 15, 2026

andrew129260 said:
Yes, despite what some people say, AI is here to stay, and this will keep happening.

It's Internet 2.0.

Ai will bring lots of good, and lots of bad.

Finding these flaws are good imho, as long as they are patched later. The bad news is cyber warfare will rise highly.

Good point, finding them is a good thing, but the downside is people using them maliciously.

And this is only the tip of the iceberg when it comes to AI. I see some possibly interesting times ahead for humanity as AI gains more of a foothold.

Here's an idea, they should let AI analyze Windows & see how many flaws it finds....

andrew129260 · May 15, 2026

Borg 386 said:
Here's an idea, they should let AI analyze Windows & see how many flaws it finds....

They actually are, google, microsoft, meta etc and a bunch of other companies were given mythos for free to scrape their code for errors, due to the concern of how effective it has already been in firefox and other things. A lot of people thought it was just marketing at first, but due to the number of vulnerabilities, anthropic was generally concerned and gave them access to the model without cost.

Microsoft statement on mythos:

Strengthening secure software at global scale: How MSRC is evolving with AI

Anthropic: Project Glasswing

Also, Google project zero has already been doing this for quite some time, but with human vulnerability testers, not AI.

Google Project Zero

Make zeroday hard

projectzero.google

Borg 386 · May 15, 2026

andrew129260 said:
They actually are, google, microsoft, meta etc and a bunch of other companies were given mythos for free to scrape their code for errors, due to the concern of how effective it has already been in firefox and other things. A lot of people thought it was just marketing at first, but due to the number of vulnerabilities, anthropic was generally concerned and gave them access to the model without cost.

Microsoft statement on mythos:

Strengthening secure software at global scale: How MSRC is evolving with AI

Anthropic: Project Glasswing

Also, Google project zero has already been doing this for quite some time, but with human vulnerability testers, not AI.

Google Project Zero

Make zeroday hard

projectzero.google

Well I wish they would hurry up & use it a bit more. I'm tired of having to drape a Rosery across my PC & saying a prayer every time I apply patches.

neemobeer · May 15, 2026

The real challenge of having AI is the average time to exploit after a patch is released. Defenders used to have 2-4 weeks or more to patch before threat actors had working exploits. AI can now develop a working PoC exploit in around 20 hours. Mythos isn't the only model that can do this. Many are publicly available. It's puts stronger emphasis on doing the fundamentals of cyber security well and have good layered defends so that patching doesn't become a companies Achilles heel

maur0 · May 15, 2026

Now all that's left is for Microsoft to use AI to test everything, find the bugs, and then release a Windows 11 V2 AI review edition without bugs and with tripled performance.
Then, who knows, maybe in 2040 that will happen?

andrew129260 · May 15, 2026

neemobeer said:
The real challenge of having AI is the average time to exploit after a patch is released. Defenders used to have 2-4 weeks or more to patch before threat actors had working exploits. AI can now develop a working PoC exploit in around 20 hours. Mythos isn't the only model that can do this. Many are publicly available. It's puts stronger emphasis on doing the fundamentals of cyber security well and have good layered defends so that patching doesn't become a companies Achilles heel

Extremely well said, thank you

XxXxX · May 15, 2026

all 3 security flaws in the Linux kernel use the same/similar exploit but also need access to that system.

copy fail fix

Fixes available for CVE-2026-31431 (Copy Fail) Linux Kernel Local Privilege Escalation Vulnerability | Ubuntu

A local privilege escalation (LPE) vulnerability affecting the Linux kernel has been publicly disclosed on April 29, 2026. The vulnerability has been assigned CVE ID CVE-2026-31431 and is referred to as Copy Fail. The affected component is a kernel module that provides hardware-accelerated...

ubuntu.com

dirty frag fix

Dirty Frag Linux kernel local privilege escalation vulnerability mitigations | Ubuntu

Two local privilege escalation (LPE) vulnerabilities affecting the Linux kernel have been publicly disclosed on May 7, 2026. The vulnerabilities have been assigned the IDs CVE-2026-43284 and CVE-2026-43500 and are referred to as “Dirty Frag.” The affected components are Linux kernel modules. The...

ubuntu.com

fragnesia fix

Fragnesia Linux kernel local privilege escalation vulnerability mitigations | Ubuntu

A local privilege escalation (LPE) vulnerability affecting the Linux kernel has been publicly disclosed on May 13, 2026. The vulnerability does not have a CVE ID published, but is referred to as “Fragnesia.” The vulnerability affects multiple Linux distributions, including all Ubuntu releases...

ubuntu.com

the 7.0 kernel has already been patched for copyfail and dirtyfrag
fragnesia will patched shortly. but fixes for all three are available.

these flaws in the kernel has been known for some time
its only now they have been made public that they are just starting to provide kernel patches
best of luck Steve ..

Alejandro85 · May 19, 2026

With such things available, properly using Responsible Disclosure becomes more critical than ever

jimbo45 · May 20, 2026

XxXxX said:
all 3 security flaws in the Linux kernel use the same/similar exploit but also need access to that system.

copy fail fix

Fixes available for CVE-2026-31431 (Copy Fail) Linux Kernel Local Privilege Escalation Vulnerability | Ubuntu

A local privilege escalation (LPE) vulnerability affecting the Linux kernel has been publicly disclosed on April 29, 2026. The vulnerability has been assigned CVE ID CVE-2026-31431 and is referred to as Copy Fail. The affected component is a kernel module that provides hardware-accelerated...

ubuntu.com

dirty frag fix

Dirty Frag Linux kernel local privilege escalation vulnerability mitigations | Ubuntu

Two local privilege escalation (LPE) vulnerabilities affecting the Linux kernel have been publicly disclosed on May 7, 2026. The vulnerabilities have been assigned the IDs CVE-2026-43284 and CVE-2026-43500 and are referred to as “Dirty Frag.” The affected components are Linux kernel modules. The...

ubuntu.com

fragnesia fix

Fragnesia Linux kernel local privilege escalation vulnerability mitigations | Ubuntu

A local privilege escalation (LPE) vulnerability affecting the Linux kernel has been publicly disclosed on May 13, 2026. The vulnerability does not have a CVE ID published, but is referred to as “Fragnesia.” The vulnerability affects multiple Linux distributions, including all Ubuntu releases...

ubuntu.com

the 7.0 kernel has already been patched for copyfail and dirtyfrag
fragnesia will patched shortly. but fixes for all three are available.

these flaws in the kernel has been known for some time
its only now they have been made public that they are just starting to provide kernel patches
best of luck Steve ..

7.1 Kernel already patched / fixed etc. As there are far more commercial cloud servers etc using Linux than Windows Server these things are usually fixed when they occur pretty quickly -- Windows commercial servers are often created as "Virtual servers" from Linux cloud systems.

Cheers
jimbo

Solved The third major Linux kernel flaw in two weeks has been found - thanks to AI

Well-known member

My Computer

Well-known member

My Computers

Well-known member

My Computer

Well-known member

My Computers

Well-known member

My Computer

Well-known member

My Computer

Well-known member

My Computer

Well-known member

My Computers

Just a Computer User

My Computers

New member

My Computer

Well-known member

My Computer