Serious Discussion How Does One improve Security ?

[Divine_Barakah]

Guys, please stay on topic. This thread is for discussing "How does one improve security'.

I shared some insights and some links about disclosed vulnerabilities that would help other better understand the threat vectors. Each one of the replies I made is one way or another related to the core of this topic.

Edit:


Correcting misinformation and dismantling unverified claims (on the topic of cyber security) is precisely on-topic.

 

[Parkinsond]

So now you resort to moving the goalposts. Each time you throw some unverified claim

You could politely explained in details the alternative point of view, the way I used to by decent members such as Andy Ful.
But you choosed the hard way to tell; you earned my way of response.

 

[Divine_Barakah]

Security News Thread '16-Year-Old Linux KVM Flaw Lets Guest VMs Escape to Host on Intel and AMD x86 Systems'

Jul 6, 2026

A use-after-free bug in Linux's KVM hypervisor can be triggered from a guest virtual machine to corrupt the shadow-page state of the host kernel that runs it.

Dubbed 'Januscape' and tracked as CVE-2026-53359, the flaw sits in the shadow MMU code that KVM shares across both Intel and AMD. The public proof-of-concept panics the host; the researcher claims that a separate, unreleased exploit turns the same bug into full host code execution.

Security researcher Hyunwoo Kim (@v4bel) found and reported the bug. He described Januscape as the first guest-to-host exploit triggerable on both...

• Captain Awesome
• Replies: 1
• Forum: Security News

• 
• 

This completely nukes the fallacy of "up-to-date"

The Januscapel Vulnerability was introduced in the Linux kernel in August 2010, and it was not patched until 2026 (16 years!)

For a decade and a half, Linux system administrators who religiously kept their servers "up-to-date" were still completely vulnerable to this zero-day exploit. An updated OS only protects against known threats; it does nothing to stop a 16-year-old architectural flaw.

 

[Victor M]

Guys, this thread is for posting Methods to Improve Security.

@Divine_Barakah has demo'ed another way, and that is to learn from attack dissections by security researchers. Some call it Open Source Intelligence. (OSINT). Members of MT frequently post these as Security News in the forum. Sometimes, the dissections reveal methods where we can block some of the attacks. For example for ClickFix attacks, we can use Group Policy to forbid use of the Run Box ( accessed by Winkey + R )

 

[stonjean633]

The tech is there. I should say User Awareness. Once people are aware of the do's and don'ts, it will help any organization be it SMB or Enterprise.

 

[Deletedmessiah]

Best way is to be rich and hire some expert to manage it for you. Easy.

 

[Acadia]

Best way is to be rich and hire some expert to manage it for you. Easy.

Easy to be rich?
Acadia

 

[Victor M]

Once people are aware of the do's and don'ts,

The best way is to write all those do's and don'ts as procedures in a procedure manual so that people can read and follow religiously. Like a laminated one or two pager as part of an employee kit.

 

[stonjean633]

The best way is to write all those do's and don'ts as procedures in a procedure manual so that people can read and follow religiously. Like a laminated one or two pager as part of an employee kit.

And that procedure goes into a Corporate Policy.
Having it this way and with Leadership approval, all must abide by it

 

[Victor M]

I found one way to improve my security is to list out all the security configurations that I make - AV and it's settings, Hard Configurator settings, all hardening steps into a document and upload it to AI, and then ask it to criticize the security, identify the security gaps and attack vectors I've missed, and suggest compensating controls that addresses the security gaps.

I have subscribed to Anthropic Claude. And asked for critique, twice. And both times the AI suggested different things. So remember to ask it at least twice.

 

[Victor M]

Remote execution requires a footstep on your device

Not true. Remote execution is just as it the term says, the adversary could issue commands over the net.

 

[Victor M]

I firmly believe in having a non-default-security set up. Having a plain Win 11 out of box and a popular AV without configuration is asking for trouble. Hackers come prepared for that sort of set up.

And the most vulnerable stage of a box is during the setup stage, when defenses are not all in place. Have a written down hardening check list, divide it into offline do-able and online to-do's, do it up to the point of needing to go online. Do a drive image, then continue. That way if something bad happens or something feels weird while online in the middle of setup, you can quickly revert and do it again. And when it does all go as planned, save a drive image too.

'Feels weird' is hard to pin down. When you have the UAC slider all the way to the top, and an expected UAC doesn't show up. That sort of defines weird. That was in the past. A recent one was that I was doing an dnf update (online update of Fedora Linux) and suddenly I get a bunch of red lines appear indicating a disruption. Network errors are not weird, but later my perfectly entered password that I have muscle memory of refuses to authenticate, a few times, I decided at the time that I'd let it slide. Watch me regret it. There is the pre-online drive image. Lets see if I have to 'start' over.

 

[Parkinsond]

Not true. Remote execution is just as it the term says, the adversary could issue commands over the net.

Need more explanation; how it can do it withouth footstep, with all inbound traffic blocked?
If you are correct, anyone can be easily RATed.

 

[Victor M]

Need more explanation;

I cannot recall the exact CVE, but for example it was vulnerability on a router, and the attacker can remotely activate a flaw, and run code.

And don't think Windows is immune from these sort of things, Windows take input from the network in many instances. For example your Windows Spotlight randomly receives 'photos' ( a glob of something). and if there's a flaw in the graphics processing, it could get hit. For one, the firewall does not expose a control over it.