Malware Research

19 August 2026 15:51

That evidence-first approach sounds right. For ambiguous cases, a short record of the evidence, alternatives considered and what would disprove the conclusion could be as useful as the final answer. It makes uncertainty visible rather than letting the agent hide it.

Malware Research

19 August 2026 12:32

That’s a good point. I’m still experimenting with different samples, from simple to more complex and ambiguous ones.
So far, I’ve mainly focused on whether the agent can gather the right evidence and reach conclusions that can be verified manually in IDA. Testing misleading routines and uncertainty handling would definitely be a useful next step.

Malware Research

19 August 2026 05:56

New FedBan
Fed: Libra's Empire
FedAdmin: ❤🦦
User: Jack Truan
User ID: 8746380384
Reason: scam

Malware Research

18 August 2026 14:57

The useful test for an IDAPython agent is not whether it names an encryption algorithm but whether every conclusion links back to a function, trace or decompiler view an analyst can inspect. That makes the tool easier to trust and to correct when it is wrong.

Malware Research

17 August 2026 19:34

Oh, didn't know Andreas Zeller created this as well. His course on debugging is great

Malware Research

17 August 2026 18:43

New FedBan
Fed: Libra's Empire
FedAdmin: alex 27
User: Margaret R Rowley
User ID: 976225967
Reason: spam

Malware Research

17 August 2026 17:31

hey guy, anyone know fuzzing binary :v i need learn ab this techniques :v

Malware Research

16 August 2026 17:57

Please provide a description

Malware Research

16 August 2026 10:50

If you would like to join the Nullspire hacking team, send us a message.

Malware Research

15 August 2026 08:15

Can someone grab this from VT for me?

006f0ba963a63d9b2822b139ac806dee71eb6a3382b1b1db74b5cf2e60b57a51

Malware Research

14 August 2026 13:43

Before looking at a sample, it helps to state what you observed and what you need to confirm. A hash alone gives little context; the initial execution chain and sandbox behaviour make the analysis question much clearer.

Malware Research

14 August 2026 01:00

The two cases I know did exactly the same thing, just spammed these four pictures

Malware Research

14 August 2026 00:52

Hello, anyone here know the malware that do this? I would like a sample to analyze or at least a report.

Malware Research

13 August 2026 22:45

Analysis of Android FvncBot banker campaign targeting Polish users
https://cert.pl/en/posts/2026/03/fvncbot-analysis/

Malware Research

13 August 2026 22:09

Phishing is a Social Engineering where someone is lured into doing something they normally wouldn't do by deceiving them using what they are familiar with. It can be applied in many things depending on what the phisherman(The Engineer) is targeting. For example login credentials, sensitive documents, Files, etc.
Most times the most used case is phishing for login credentials. In a recent Phishing Pentesting engagement I and my team carried out on the financial department of a firm was centered on login (Access) where we used Chase Bank as a case study to train them on how to spot a phishing attack. The tool we designed for it is made to phish login credential and OTP and deliver them to a telegram account in real time just as an Attacker would.
If you want to use it to learn more about phishing and how to defend yourself against it, I can give you the code.

Malware Research

19 August 2026 13:06

New FedBan
Fed: Libra's Empire
FedAdmin: alex 27
User: @​Co4de
User ID: 7724068664
Reason: spam

Malware Research

19 August 2026 12:19

Matching a manual walkthrough is a strong sanity check. The next useful test would be a deliberately misleading sample or ambiguous routine, because that shows whether the agent can express uncertainty instead of forcing a clean answer. Does the workflow keep a record of rejected hypotheses too?

Malware Research

18 August 2026 16:59

Absolutely, I agree. That is one reason I find integrating the agent with IDA useful. The goal is not just for the agent to give a conclusion, but to link it back to the relevant function, decompiled code, cross-references, or other evidence.
The analyst can then verify the conclusion directly in IDA and correct the agent if needed. I think that kind of traceability and human verification is very important when using AI for reverse engineering.
For reference, here is a separate video where I manually analyze how the malware encrypts the captured keystrokes. The findings from the manual analysis match what the AI agent identified in the demo:
https://youtu.be/EgOQsfhb_j0

Malware Research

18 August 2026 14:28

I’ve been experimenting with how AI agents can assist with malware reverse engineering by interacting directly with reverse engineering tools and helping analysts investigate unfamiliar code.
In this video, watch an AI agent use tools exposed through IDAPython to analyze a malware sample and determine the encryption algorithm and encryption key used to encrypt captured keystrokes before they are stored in a file. This is currently a quick demonstration without detailed audio narration. I’ll be creating a more detailed video with audio soon, where I’ll walk through the setup, tools, and the complete reverse engineering workflow step by step.
For now, have a look at the demo to see how an AI agent can reason over disassembled code, leverage IDA capabilities through IDAPython, and assist with extracting meaningful information from malware.
AI-Powered Reverse Engineering:
https://youtu.be/gK9BFd_5OAc

Malware Research

17 August 2026 19:28

https://www.fuzzingbook.org/

This Is a good resource

Malware Research

17 August 2026 18:10

Check open security training v2 web site

Malware Research

17 August 2026 06:46

New FedBan
Fed: Libra's Empire
FedAdmin: ❤🦦
User: Eleanor E werner
User ID: 810745975
Reason: spam

Malware Research

16 August 2026 15:20

New FedBan
Fed: Libra's Empire
FedAdmin: ❤🦦
User: Nullspire
User ID: 8013035498
Reason: illegal activity

Malware Research

15 August 2026 15:56

That makes the reporting task clearer. First get the expected behaviours and labels from the team, then use curated samples in an isolated lab and record what the sandbox observes rather than only its verdict. Different platforms can disagree because their telemetry, signatures and execution environment differ. Your report should show those gaps instead of treating one verdict as ground truth.

Malware Research

14 August 2026 15:40

New FedBan
Fed: Libra's Empire
FedAdmin: alex 27
User: Quarix
User ID: 8455952624
Reason: no maldev

Malware Research

14 August 2026 07:12

File parsers are a nice starting point because a crashing input is easy to keep with the project. Save the parser version and exact command next to every crash so later changes do not turn it into a ghost bug. Which format are you using first?

Malware Research

14 August 2026 00:54

I’ve seen two discord users already spamming this in the last two weeks

Malware Research

14 August 2026 00:49

For file parser, actually I making a project on this- fuzzing-lab

Malware Research

13 August 2026 22:17

That is not true. Though there are malicious codes that do such a but it is not just from downloading a photo. Codes that can do such things are Remote Administrator Tools (RATs) but modern devices has been engineered to resist such attacks though not completely as a well tailored Social Engineering campaign can still succeed.

Malware Research

13 August 2026 09:40

Anyone from SOC background here?