input stringlengths 1 785 | output dict | schema listlengths 0 0 |
|---|---|---|
The RSA-2048 public key to verify integrity of the executable inside the GIF file is the same as in the previous version of MiniDuke. | {
"entities": {
"entity": [
{
"text": "The RSA-2048 public key",
"start": 0,
"end": 23
},
{
"text": "integrity of the executable inside the GIF file",
"start": 34,
"end": 81
}
],
"action": [
{
"text": "verify",
"... | [] |
In the event that MiniDuke is unable to retrieve a C & C URL from this account, it generates a username to search for based on the current date. | {
"entities": {}
} | [] |
The search query changes roughly every seven days and is similar to the backup mechanism in previous versions that was using Google searches. | {
"entities": {}
} | [] |
A Python implementation of the algorithm can be found in Appendix B. | {
"entities": {}
} | [] |
The TwitterJS module is extracted by creating a copy of the Windows DLL cryptdll.dll, injecting a block of code into it and redirecting the exported functions to this code. | {
"entities": {
"action": [
{
"text": "creating",
"start": 37,
"end": 45
}
],
"entity": [
{
"text": "a copy of the Windows DLL cryptdll.dll",
"start": 46,
"end": 84
}
]
}
} | [] |
Here is how the export address table of the patched binary looks after modifications. | {
"entities": {}
} | [] |
This file is then stored in an Alternate Data Stream (ADS) in NTUSER.DAT in the% USERPROFILE% folder. | {
"entities": {
"entity": [
{
"text": "This file",
"start": 0,
"end": 9
},
{
"text": "an Alternate Data Stream (ADS) in NTUSER.DAT in the% USERPROFILE% folder",
"start": 28,
"end": 100
}
],
"action": [
{
"text": "is then... | [] |
Finally this DLL is registered as the Open command when a drive is open, which has the effect of starting the bot every time the user opens a disk drive. | {
"entities": {
"entity": [
{
"text": "this DLL",
"start": 8,
"end": 16
},
{
"text": "the Open command when a drive is open",
"start": 34,
"end": 71
},
{
"text": "the bot",
"start": 106,
"end": 113
},
... | [] |
Below you can find the content of the init.cmd script used by MiniDuke to install TwitterJS module onto the system. | {
"entities": {
"entity": [
{
"text": "MiniDuke",
"start": 62,
"end": 70
},
{
"text": "TwitterJS module",
"start": 82,
"end": 98
},
{
"text": "the system",
"start": 104,
"end": 114
}
],
"action": [
... | [] |
When loaded, TwitterJS instantiates the JScript COM object and decrypts a JScript file containing the core logic of the module. | {
"entities": {}
} | [] |
Prior to executing it, MiniDuke applies a light encoding to the script: The next images show the result of two separate obfuscations, we can see that the variables have different values. | {
"entities": {
"entity": [
{
"text": "MiniDuke",
"start": 23,
"end": 31
},
{
"text": "a light encoding",
"start": 40,
"end": 56
},
{
"text": "the script",
"start": 60,
"end": 70
}
],
"action": [
... | [] |
This is probably done to thwart security systems that scan at the entry points of the JScript engine. | {
"entities": {
"entity": [
{
"text": "This",
"start": 0,
"end": 4
},
{
"text": "security systems that scan at the entry points of the JScript engine",
"start": 32,
"end": 100
}
],
"action": [
{
"text": "thwart",
... | [] |
The purpose of this script is to use Twitter to find a C & C and retrieve JScript code to execute. | {
"entities": {
"entity": [
{
"text": "this script",
"start": 15,
"end": 26
},
{
"text": "Twitter",
"start": 37,
"end": 44
},
{
"text": "find a C & C",
"start": 48,
"end": 60
},
{
"text": "JSc... | [] |
It first generates a Twitter user to search for; this search term changes every 7 days and is actually a match to the real account name, not the Twitter account name. | {
"entities": {}
} | [] |
The bot then visits the Twitter profiles returned by the search and looks for links that end with'' .xhtml | {
"entities": {}
} | [] |
''. | {
"entities": {}
} | [] |
When one is found, it replaces'' .xhtml | {
"entities": {}
} | [] |
'' with'' .php | {
"entities": {}
} | [] |
'' and fetches that link. | {
"entities": {}
} | [] |
Information about the computer is embedded in the Accept HTTP header. | {
"entities": {}
} | [] |
The first link on the retrieved page should contain base64 data; the name attribute of the link is used as a rolling XOR key to decrypt the JScript code. | {
"entities": {}
} | [] |
Finally, MiniDuke calculates a hash of the fetched script and compares it with a hardcoded hash in the TwitterJS script. | {
"entities": {}
} | [] |
If they match, the fetched script is executed by calling eval (). | {
"entities": {}
} | [] |
The code hashing algorithm used by the component looks very much like SHA-1 but outputs different hashes (you can find the complete implementation in Appendix B. | {
"entities": {}
} | [] |
We decided to search for what was changed in the algorithm; one of our working hypotheses was that the algorithm might have been altered to make collisions feasible. | {
"entities": {}
} | [] |
We couldn't find an obvious difference; all the constants and the steps of the algorithm were as expected. | {
"entities": {}
} | [] |
Then we noticed that for short messages only the second 32-bit word was different when compared to the original SHA-1. | {
"entities": {}
} | [] |
By examining how this 2nd word was generated we finally discovered that this was caused by a scope issue. | {
"entities": {}
} | [] |
As shown below the SHA-1 function used a variable named f: the function Z () is then called which also uses a variable named f without the var keyword, causing it to be treated as a global variable rather than local to the function. | {
"entities": {}
} | [] |
The end result is that the value of f is also changed in the SHA-1 function which affects the value of the 2nd word for that round and ultimately the whole hash for long messages. | {
"entities": {}
} | [] |
A likely explanation of how this problem came to be is that the variable names were changed to single letters using an automated tool prior to embedding it in the payload. | {
"entities": {}
} | [] |
The 2 f variables probably had different names in the original script which avoided the issue. | {
"entities": {}
} | [] |
So this leaves us with two takeaways: 1) The difference in the hashing algorithm was unintentional and 2) Always declare your local variables with the var keyword. | {
"entities": {}
} | [] |
;-) We generated the list of Twitter search terms for 2013-2014 and checked if any of those were registered. | {
"entities": {}
} | [] |
At the moment only one exists, @AA2ADcAOAA, which is the TwitterJS account that was generated between August 21st and 27th 2013. | {
"entities": {}
} | [] |
This account has no tweets. | {
"entities": {}
} | [] |
In an effort to discover potential victims, we registered the Twitter accounts corresponding to the current week both for the main and TwitterJS components and set up tweets with encrypted URLs so that an infected computer would reach out to our server. | {
"entities": {}
} | [] |
So far we have received connections via the TwitterJS accounts from four computers located in Belgium, France and the UK. | {
"entities": {}
} | [] |
We have contacted national CERTs to notify the affected parties. | {
"entities": {}
} | [] |
We detect the RTF exploit document as Win32/Exploit.CVE-2014-1761.D and the MiniDuke components as Win32/SandyEva.G. | {
"entities": {}
} | [] |
In the world of malware threats, only a few rare examples can truly be considered groundbreaking and almost peerless. | {
"entities": {}
} | [] |
What we have seen in Regin is just such a class of malware. | {
"entities": {}
} | [] |
Regin is an extremely complex piece of software that can be customized with a wide range of different capabilities which can be deployed depending on the target. | {
"entities": {
"entity": [
{
"text": "Regin",
"start": 0,
"end": 5
}
]
}
} | [] |
It is built on a framework that is designed to sustain long-term intelligence-gathering operations by remaining under the radar. | {
"entities": {
"entity": [
{
"text": "It",
"start": 0,
"end": 2
},
{
"text": "long-term intelligence-gathering operations",
"start": 55,
"end": 98
},
{
"text": "the radar",
"start": 118,
"end": 127
}
]... | [] |
It goes to extraordinary lengths to conceal itself and its activities on compromised computers. | {
"entities": {
"entity": [
{
"text": "It",
"start": 0,
"end": 2
},
{
"text": "itself and its activities on compromised computers",
"start": 44,
"end": 94
}
],
"action": [
{
"text": "conceal",
"start": 36,
... | [] |
Its stealth combines many of the most advanced techniques that we have ever seen in use. | {
"entities": {}
} | [] |
The main purpose of Regin is intelligence gathering and it has been implicated in data collection operations against government organizations, infrastructure operators, businesses, academics, and private individuals. | {
"entities": {
"entity": [
{
"text": "The main purpose of Regin",
"start": 0,
"end": 25
},
{
"text": "intelligence gathering",
"start": 29,
"end": 51
},
{
"text": "it",
"start": 56,
"end": 58
},
{
... | [] |
The level of sophistication and complexity of Regin suggests that the development of this threat could have taken well-resourced teams of developers many months or years to develop and maintain. | {
"entities": {}
} | [] |
Regin is a multi-staged, modular threat, meaning that it has a number of components, each depending on others, to perform attack operations. | {
"entities": {}
} | [] |
This modular approach gives flexibility to the threat operators as they can load custom features tailored to individual targets when required. | {
"entities": {}
} | [] |
Some custom payloads are very advanced and exhibit a high degree of expertise in specialist sectors. | {
"entities": {}
} | [] |
The modular design also makes analysis of the threat difficult, as all components must be available in order to fully understand it. | {
"entities": {
"entity": [
{
"text": "The modular design",
"start": 0,
"end": 18
},
{
"text": "analysis of the threat difficult",
"start": 30,
"end": 62
}
],
"action": [
{
"text": "makes",
"start": 24,
"... | [] |
This modular approach has been seen in other sophisticated malware families such as Flamer and Weevil (The Mask), while the multi-stage loading architecture is similar to that seen in the Duqu/Stuxnet family of threats. | {
"entities": {}
} | [] |
Regin is different to what are commonly referred to as'' traditional'' advanced persistent threats (APTs), both in its techniques and ultimate purpose. | {
"entities": {}
} | [] |
APTs typically seek specific information, usually intellectual property. | {
"entities": {
"entity": [
{
"text": "APTs",
"start": 0,
"end": 4
},
{
"text": "specific information, usually intellectual property",
"start": 20,
"end": 71
}
],
"action": [
{
"text": "seek",
"start": 15,
... | [] |
Regin's purpose is different. | {
"entities": {}
} | [] |
It is used for the collection of data and continuous monitoring of targeted organizations or individuals. | {
"entities": {
"entity": [
{
"text": "It",
"start": 0,
"end": 2
},
{
"text": "the collection of data",
"start": 15,
"end": 37
},
{
"text": "targeted organizations or individuals",
"start": 67,
"end": 104
}... | [] |
This report provides a technical analysis of Regin based on a number of identified samples and components. | {
"entities": {}
} | [] |
This analysis illustrates Regin's architecture and the many payloads at its disposal. | {
"entities": {}
} | [] |
Regin is a multi-purpose data collection tool which dates back several years. | {
"entities": {}
} | [] |
Symantec first began looking into this threat in the fall of 2013. | {
"entities": {}
} | [] |
Multiple versions of Regin were found in the wild, targeting several corporations, institutions, academics, and individuals. | {
"entities": {}
} | [] |
Regin has a wide range of standard capabilities, particularly around monitoring targets and stealing data. | {
"entities": {
"entity": [
{
"text": "Regin",
"start": 0,
"end": 5
},
{
"text": "targets",
"start": 80,
"end": 87
},
{
"text": "data",
"start": 101,
"end": 105
}
],
"action": [
{
"tex... | [] |
It also has the ability to load custom features tailored to individual targets. | {
"entities": {}
} | [] |
Some of Regin's custom payloads point to a high level of specialist knowledge in particular sectors, such as telecoms infrastructure software, on the part of the developers. | {
"entities": {}
} | [] |
Regin is capable of installing a large number of additional payloads, some highly customized for the targeted computer. | {
"entities": {
"entity": [
{
"text": "Regin",
"start": 0,
"end": 5
},
{
"text": "a large number of additional payloads",
"start": 31,
"end": 68
}
],
"action": [
{
"text": "installing",
"start": 20,
"end"... | [] |
The threat's standard capabilities include several remote access Trojan (RAT) features, such as capturing screenshots and taking control of the mouse's point-and-click functions. | {
"entities": {
"entity": [
{
"text": "The threat's standard capabilities",
"start": 0,
"end": 34
},
{
"text": "several remote access Trojan (RAT) features",
"start": 43,
"end": 86
},
{
"text": "screenshots",
"start": 10... | [] |
Regin is also configured to steal passwords, monitor network traffic, and gather information on processes and memory utilization. | {
"entities": {
"entity": [
{
"text": "Regin",
"start": 0,
"end": 5
},
{
"text": "passwords",
"start": 34,
"end": 43
},
{
"text": "network traffic",
"start": 53,
"end": 68
},
{
"text": "inform... | [] |
It can also scan for deleted files on an infected computer and retrieve them. | {
"entities": {
"entity": [
{
"text": "It",
"start": 0,
"end": 2
},
{
"text": "deleted files on an infected computer",
"start": 21,
"end": 58
},
{
"text": "them",
"start": 72,
"end": 76
}
],
"action... | [] |
More advanced payload modules designed with specific goals in mind were also found in our investigations. | {
"entities": {}
} | [] |
For example, one module was designed to monitor network traffic to Microsoft Internet Information Services (IIS) web servers, another was designed to collect administration traffic for mobile telephony base station controllers, while another was created specifically for parsing mail from Exchange databases. | {
"entities": {
"entity": [
{
"text": "one module",
"start": 13,
"end": 23
},
{
"text": "network traffic to Microsoft Internet Information Services (IIS) web servers",
"start": 48,
"end": 124
},
{
"text": "another",
"sta... | [] |
Regin goes to some lengths to hide the data it is stealing. | {
"entities": {
"entity": [
{
"text": "Regin",
"start": 0,
"end": 5
},
{
"text": "the data",
"start": 35,
"end": 43
},
{
"text": "it",
"start": 44,
"end": 46
}
],
"action": [
{
"text":... | [] |
Valuable target data is often not written to disk. | {
"entities": {
"entity": [
{
"text": "Valuable target data",
"start": 0,
"end": 20
},
{
"text": "disk",
"start": 45,
"end": 49
}
],
"action": [
{
"text": "is often not written",
"start": 21,
"end": 41
... | [] |
In some cases, Symantec was only able to retrieve the threat samples but not the files containing stolen data. | {
"entities": {}
} | [] |
Symantec is aware of two distinct versions of Regin. | {
"entities": {}
} | [] |
Version 1.0 appears to have been used from at least 2008 to 2011. | {
"entities": {}
} | [] |
Version 2.0 has been used from 2013 onwards, though it may have possibly been used earlier. | {
"entities": {}
} | [] |
Version 1.0 appears to have been abruptly withdrawn from circulation in 2011. | {
"entities": {}
} | [] |
Version 1.0 samples found after this date seem to have been improperly removed or were no longer accessible to the attackers for removal. | {
"entities": {}
} | [] |
This report is based primarily on our analysis of Regin version 1.0. | {
"entities": {}
} | [] |
We also touch on version 2.0, for which we only recovered 64-bit files. | {
"entities": {}
} | [] |
Symantec has assigned these version identifiers as they are the only two versions that have been acquired. | {
"entities": {}
} | [] |
Regin likely has more than two versions. | {
"entities": {}
} | [] |
There may be versions prior to 1.0 and versions between 1.0 and 2.0. | {
"entities": {}
} | [] |
The Regin operators do not appear to focus on any specific industry sector. | {
"entities": {}
} | [] |
Regin infections have been observed in a variety of organizations, including private companies, government entities, and research institutes. | {
"entities": {}
} | [] |
Infections are also geographically diverse, having been identified mainly in 10 different regions. | {
"entities": {}
} | [] |
The infection vector varies among targets. | {
"entities": {}
} | [] |
A reproducible infection vector is unconfirmed at time of writing. | {
"entities": {}
} | [] |
Targets may be tricked into visiting spoofed versions of well- known websites and the threat may be installed through a web browser or by exploiting an application. | {
"entities": {
"entity": [
{
"text": "the threat",
"start": 82,
"end": 92
},
{
"text": "a web browser",
"start": 118,
"end": 131
},
{
"text": "exploiting an application",
"start": 138,
"end": 163
}
],
... | [] |
On one computer, log files show that Regin originated from Yahoo! Instant Messenger through an unconfirmed exploit. | {
"entities": {
"entity": [
{
"text": "Regin",
"start": 37,
"end": 42
},
{
"text": "Yahoo! Instant Messenger",
"start": 59,
"end": 83
},
{
"text": "an unconfirmed exploit",
"start": 92,
"end": 114
}
],
... | [] |
Regin has a six-stage architecture. | {
"entities": {}
} | [] |
The initial stages involve the installation and configuration of the threat's internal services. | {
"entities": {
"entity": [
{
"text": "The initial stages",
"start": 0,
"end": 18
},
{
"text": "the installation and configuration of the threat's internal services",
"start": 27,
"end": 95
}
],
"action": [
{
"text": "in... | [] |
The later stages bring Regin's main payloads into play. | {
"entities": {}
} | [] |
This section presents a brief overview of the format and purpose of each stage. | {
"entities": {}
} | [] |
The most interesting stages are the executables and data files stored in Stages 4 and 5. | {
"entities": {}
} | [] |
The initial Stage 1 driver is the only plainly visible code on the computer. | {
"entities": {}
} | [] |
All other stages are stored as encrypted data blobs, as a file or within a non-traditional file storage area such as the registry, extended attributes, or raw sectors at the end of disk. | {
"entities": {
"entity": [
{
"text": "All other stages",
"start": 0,
"end": 16
},
{
"text": "encrypted data blobs",
"start": 31,
"end": 51
}
],
"action": [
{
"text": "are stored",
"start": 17,
"end": 27
... | [] |
Symantec Security Response has not obtained the Regin dropper at the time of writing. | {
"entities": {}
} | [] |
Symantec believes that once the dropper is executed on the target's computer, it will install and execute Stage 1. | {
"entities": {
"entity": [
{
"text": "the dropper",
"start": 28,
"end": 39
},
{
"text": "it",
"start": 78,
"end": 80
},
{
"text": "Stage 1",
"start": 106,
"end": 113
}
],
"action": [
{
... | [] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.