Software Review The biggest risk with Windows: LOLBINS

[ForgottenSeer 114717]

There is very good AI available. Users need to start using it to learn how to do robust security.

 

[simmerskool]

There is very good AI available. Users need to start using it to learn how to do robust security.

I learning some Python fwiw with chatGPT. my last coding experience was fortran circa 1968. having fun, but not really security related, unless I'm opening up holes

 

[Trident]

I learning some Python fwiw with chatGPT. my last coding experience was fortran circa 1968. having fun, but not really security related, unless I'm opening up holes

Love Fortran and any language that doesn't step on a runtime generally. With Fortran the syntax WAS so intuitive until it became somewhat a spin-off of C++ (not that there is something wrong).

 

[Jonny Quest]

There is very good AI available. Users need to start using it to learn how to do robust security.

I know Gemini isn't as robust as some of what you all use, but it has been very helpful with my some of my security questions.

Including it pointing out that in using Glasswire free along with F- Secure, even though it was showing first outbound connections, like I wanted it to, if I were to get something like an infostealer that when that notification was showing, it was already to late, the outbound connection had already been made.

I switched to WFC to verify pop up notifications of connections that can then be allowed or blocked.

 

[rashmi]

There is very good AI available. Users need to start using it to learn how to do robust security.

Sure... MalwareTips threads are the ultimate testing ground—where alpha learning versions go to level up!

 

[ForgottenSeer 123960]

I know Gemini isn't as robust as some of what you all use, but it has been very helpful with my some of my security questions.

Including it pointing out that in using Glasswire free along with F- Secure, even though it was showing first outbound connections, like I wanted it to, if I were to get something like an info stealer that when that notification was showing, it was already to late, the outbound connection had already been made.

I switched to WFC to verify pop up notifications of connections that can then be allowed or blocked.

Algorithmic & Mathematical Stability
Gemini 3 Pro is currently a powerhouse here. It recently scored 100% on the AIME 2025 benchmark (with code execution), matching or beating GPT-5.2. If you need robust logical deduction or algorithmic puzzle solving, Gemini is arguably the most robust option available.

Coding & Engineering
Claude Opus 4.5 is widely considered the most "robust" for software engineering (e.g., SWE-bench Verified). It is less likely to introduce bugs during refactoring compared to Gemini or GPT. Gemini 3 is faster and excellent for "greenfield" (new) code generation, but Claude is preferred for maintaining existing complex codebases.

 

[Victor M]

Here is an article by AppGuard: Control OS Utilities on Your Endpoints before Adversaries Hurt You with Them - AppGuard . It says it can handle LoLbins and Conditionally block them.