input stringlengths 1 785 | output dict | schema listlengths 0 0 |
|---|---|---|
Some of APT28's more commonly used tools are the SOURFACE downloader, its second stage backdoor EVILTOSS, and a modular family of implants that we call CHOPSTICK. | {
"entities": {
"entity": [
{
"text": "Some of APT28's more commonly used tools",
"start": 0,
"end": 40
},
{
"text": "the SOURFACE downloader",
"start": 45,
"end": 68
},
{
"text": "its second stage backdoor EVILTOSS",
"s... | [] |
• SOURFACE: This downloader is typically called Sofacy within the cyber security community. | {
"entities": {}
} | [] |
However because we have observed the name'' Sofacy'' used to refer to APT28 malware generally (to include the SOURFACE dropper, EVILTOSS, CHOPSTICK, and the credential harvester OLDBAIT), we are using the name SOURFACE to precisely refer to a specific downloader. | {
"entities": {
"entity": [
{
"text": "the name SOURFACE",
"start": 201,
"end": 218
},
{
"text": "a specific downloader",
"start": 241,
"end": 262
}
],
"action": [
{
"text": "refer",
"start": 232,
"end": ... | [] |
This downloader obtains a second-stage backdoor from a C2 server. | {
"entities": {
"entity": [
{
"text": "This downloader",
"start": 0,
"end": 15
},
{
"text": "a second-stage backdoor",
"start": 24,
"end": 47
},
{
"text": "a C2 server",
"start": 53,
"end": 64
}
],
... | [] |
CORESHELL is an updated version of SOURFACE. | {
"entities": {}
} | [] |
• EVILTOSS: This backdoor has been delivered through the SOURFACE downloader to gain system access for reconnaissance, monitoring, credential theft, and shellcode execution. | {
"entities": {
"entity": [
{
"text": "This backdoor",
"start": 12,
"end": 25
},
{
"text": "the SOURFACE downloader",
"start": 53,
"end": 76
},
{
"text": "system access",
"start": 85,
"end": 98
},
{
... | [] |
• CHOPSTICK: This is a modular implant compiled from a software framework that provides tailored functionality and flexibility. | {
"entities": {}
} | [] |
A number of the malware variants that we profile below, especially the CHOPSTICK family, demonstrate formal coding practices indicative of methodical, diligent programmers. | {
"entities": {}
} | [] |
The modularity of CHOPSTICK alone, with its flexible and lasting platform, demonstrates planning for long-term use and versatility. | {
"entities": {}
} | [] |
We have also noted that APT28 tailors implants to their target environments, configuring them to use local network resources such as email servers. | {
"entities": {
"entity": [
{
"text": "them",
"start": 89,
"end": 93
},
{
"text": "local network resources such as email servers",
"start": 101,
"end": 146
}
],
"action": [
{
"text": "use",
"start": 97,
"... | [] |
APT28 has attempted to obfuscate their code and implement counter-analysis techniques: • One of the latest samples of CORESHELL includes counter-reverse engineering tactics via unused machine instructions. | {
"entities": {
"entity": [
{
"text": "APT28",
"start": 0,
"end": 5
},
{
"text": "their code",
"start": 33,
"end": 43
},
{
"text": "counter-analysis techniques",
"start": 58,
"end": 85
},
{
"t... | [] |
This would hinder static analysis of CORESHELL behavior by creating a large amount of unnecessary noise in the disassembly. | {
"entities": {
"entity": [
{
"text": "This",
"start": 0,
"end": 4
},
{
"text": "static analysis of CORESHELL behavior",
"start": 18,
"end": 55
},
{
"text": "creating a large amount of unnecessary noise in the disassembly",
... | [] |
• A number of CORESHELL droppers also conduct runtime checks, attempting to determine if they are executing in an analysis environment, and if so, they do not trigger their payloads. | {
"entities": {
"entity": [
{
"text": "A number of CORESHELL droppers",
"start": 2,
"end": 32
},
{
"text": "runtime checks",
"start": 46,
"end": 60
},
{
"text": "if they are executing in an analysis environment",
"start"... | [] |
• Many samples across the SOURFACE/ CORESHELL, CHOPSTICK, and EVILTOSS malware families obfuscate strings that are decoded at runtime. | {
"entities": {
"entity": [
{
"text": "Many samples across the SOURFACE/ CORESHELL, CHOPSTICK, and EVILTOSS malware families",
"start": 2,
"end": 87
},
{
"text": "strings that are decoded at runtime",
"start": 98,
"end": 133
}
],
"act... | [] |
Two of the malware families (SOURFACE/CORESHELL and EVILTOSS) use the same decryption sequence and similar algorithms for string encoding and decoding. | {
"entities": {
"entity": [
{
"text": "Two of the malware families (SOURFACE/CORESHELL and EVILTOSS)",
"start": 0,
"end": 61
},
{
"text": "the same decryption sequence and similar algorithms",
"start": 66,
"end": 117
},
{
"text"... | [] |
These families encode their strings at compile time using a custom stream cipher. | {
"entities": {
"entity": [
{
"text": "These families",
"start": 0,
"end": 14
},
{
"text": "their strings",
"start": 22,
"end": 35
},
{
"text": "compile time",
"start": 39,
"end": 51
},
{
"tex... | [] |
From a high level, these ciphers share a similar design across the malware families but differ slightly in the internal arithmetic operations. | {
"entities": {}
} | [] |
• APT28 has employed RSA encryption to protect files and stolen information moved from the victim's network to the controller. | {
"entities": {
"entity": [
{
"text": "APT28",
"start": 2,
"end": 7
},
{
"text": "RSA encryption",
"start": 21,
"end": 35
},
{
"text": "protect files and stolen information moved from the victim's network to the controller",
... | [] |
APT28 has made incremental and systematic changes to the SOURFACE downloader and its surrounding ecosystem since as early as 2007. | {
"entities": {}
} | [] |
These changes indicate a long-standing and dedicated development effort behind APT28. | {
"entities": {}
} | [] |
We have observed samples of the SOURFACE downloader compiled between 2007 and 2014. | {
"entities": {}
} | [] |
We call SOURFACE (samples are frequently named netids.dll) a first stage downloader because its primary job is to retrieve a second stage payload from a C2 server. | {
"entities": {
"entity": [
{
"text": "SOURFACE",
"start": 8,
"end": 16
},
{
"text": "a second stage payload",
"start": 123,
"end": 145
},
{
"text": "a C2 server",
"start": 151,
"end": 162
}
],
"act... | [] |
Until 2013, the SOURFACE downloader used hard-coded IP addresses for C2 communications, whereas the future CORESHELL samples use domains. | {
"entities": {
"entity": [
{
"text": "the SOURFACE downloader",
"start": 12,
"end": 35
},
{
"text": "hard-coded IP addresses",
"start": 41,
"end": 64
},
{
"text": "C2 communications",
"start": 69,
"end": 86
... | [] |
First, a malware family is a collection of malware in which each sample shares a significant amount of code with all of the others. | {
"entities": {}
} | [] |
There are exceptions: for example, some files contain public and standard code libraries that we do not take into consideration when making a family determination. | {
"entities": {}
} | [] |
A malware ecosystem is a group of malware families that work together to perform the same objective. | {
"entities": {}
} | [] |
Perhaps the simplest and most typical ecosystem is a dropper and a backdoor that are used together. | {
"entities": {}
} | [] |
They may not share the same code structure, but they are related because one drops and installs the other. | {
"entities": {
"entity": [
{
"text": "one",
"start": 73,
"end": 76
},
{
"text": "the other",
"start": 96,
"end": 105
}
],
"action": [
{
"text": "drops and installs",
"start": 77,
"end": 95
}
]
... | [] |
The ecosystem surrounding the SOURFACE downloader frequently consists of a dropper, which installs SOURFACE. | {
"entities": {
"entity": [
{
"text": "a dropper",
"start": 73,
"end": 82
},
{
"text": "SOURFACE",
"start": 99,
"end": 107
}
],
"action": [
{
"text": "installs",
"start": 90,
"end": 98
}
]
}
} | [] |
The SOURFACE downloader then receives another dropper from its C2 server, and this second dropper installs a second stage backdoor, which is usually EVILTOSS. | {
"entities": {
"entity": [
{
"text": "The SOURFACE downloader",
"start": 0,
"end": 23
},
{
"text": "another dropper",
"start": 38,
"end": 53
},
{
"text": "its C2 server",
"start": 59,
"end": 72
},
{
... | [] |
In April 2013, based on compile time, the group began to make significant alterations to the SOURFACE downloader. | {
"entities": {}
} | [] |
They started by changing the compiled DLL name to'' coreshell.dll'' and making minor changes to the network communications, as seen in Figure 7. | {
"entities": {}
} | [] |
The hostname, volume serial number and OS version data are encoded in the new URL format. | {
"entities": {}
} | [] |
As seen in the table below, the SOURFACE/ CORESHELL developers also made other modifications that changed the exported function name and file size over time. | {
"entities": {}
} | [] |
In April 2013, based on compile time, the group began to make significant alterations to the SOURFACE downloader. | {
"entities": {}
} | [] |
Variants of the SOURFACE second stage backdoor, EVILTOSS, share some code similarities with SOURFACE. | {
"entities": {}
} | [] |
However, it contains more capabilities, including the ability to provide access to the file system and registry, enumerate network resources, create processes, log keystrokes, access stored credentials, and execute shellcode. | {
"entities": {
"entity": [
{
"text": "it",
"start": 9,
"end": 11
},
{
"text": "access to the file system and registry",
"start": 73,
"end": 111
},
{
"text": "network resources",
"start": 123,
"end": 140
},... | [] |
The backdoor encrypts data that it uploads with an RSA public key. | {
"entities": {
"entity": [
{
"text": "The backdoor",
"start": 0,
"end": 12
},
{
"text": "data",
"start": 22,
"end": 26
},
{
"text": "it",
"start": 32,
"end": 34
},
{
"text": "an RSA public ke... | [] |
Many of its variants we have seen are named netui.dll. | {
"entities": {}
} | [] |
EVILTOSS variants may use the Simple Mail Transfer Protocol (SMTP) to send stolen data in an attachment named'' detaluri. dat''. | {
"entities": {
"entity": [
{
"text": "EVILTOSS variants",
"start": 0,
"end": 17
},
{
"text": "the Simple Mail Transfer Protocol (SMTP)",
"start": 26,
"end": 66
},
{
"text": "send stolen data in an attachment named'' detaluri. d... | [] |
The backdoor attaches this file to a preformatted email and sends it out through a victim's mail server. | {
"entities": {
"entity": [
{
"text": "The backdoor",
"start": 0,
"end": 12
},
{
"text": "this file",
"start": 22,
"end": 31
},
{
"text": "a preformatted email",
"start": 35,
"end": 55
},
{
"t... | [] |
Interestingly, we found an antivirus report from 2004 detailing what appears to be an early variant of EVILTOSS. | {
"entities": {}
} | [] |
The backdoor was installed alongside the NATO-themed decoy document depicted in Figure 8. | {
"entities": {
"entity": [
{
"text": "The backdoor",
"start": 0,
"end": 12
},
{
"text": "the NATO-themed decoy document depicted in Figure 8",
"start": 37,
"end": 88
}
],
"action": [
{
"text": "was installed",
"... | [] |
The backdoor sent data via SMTP to natosmtp @mail [.] ru and received its tasking via POP from natopop @mail [.] ru. | {
"entities": {
"entity": [
{
"text": "The backdoor",
"start": 0,
"end": 12
},
{
"text": "data",
"start": 18,
"end": 22
},
{
"text": "SMTP",
"start": 27,
"end": 31
},
{
"text": "natosmtp @mail... | [] |
Although we have not conclusively attributed this sample to APT28, it does suggest the possibility that APT28 has been operating since as early as 2004. | {
"entities": {}
} | [] |
A modular development framework suggests the group has had an organized development effort since as early as 2007. | {
"entities": {}
} | [] |
During our research, we discovered that APT28 uses a backdoor developed using a modular framework. | {
"entities": {
"entity": [
{
"text": "APT28",
"start": 40,
"end": 45
},
{
"text": "a backdoor developed using a modular framework",
"start": 51,
"end": 97
}
],
"action": [
{
"text": "uses",
"start": 46,
... | [] |
We call this backdoor CHOPSTICK, a somewhat ironic name that comes from our semi-random name generator. | {
"entities": {}
} | [] |
The modular design allows flexible options for compiling variants with different capabilities as needed, as well as deploying additional capabilities at runtime. | {
"entities": {}
} | [] |
This allows the developers to make targeted implants, including only the capabilities and protocols necessary for a specific environment. | {
"entities": {}
} | [] |
Such a modular framework suggests the group has had an organized development effort since as early as 2007. | {
"entities": {}
} | [] |
A formal development environment, in which code is versioned and well-organized, would almost certainly be required to track and define the various modules that can be included in the backdoor at compile time. | {
"entities": {}
} | [] |
CHOPSTICK variants may move messages and information using at least three methods: 1. | {
"entities": {}
} | [] |
Communications with a C2 server using HTTP. | {
"entities": {
"entity": [
{
"text": "Communications with a C2 server",
"start": 0,
"end": 31
},
{
"text": "HTTP",
"start": 38,
"end": 42
}
],
"action": [
{
"text": "using",
"start": 32,
"end": 37
... | [] |
These protocols are covered in more detail in Appendix D. 2. | {
"entities": {}
} | [] |
Email sent through a specified mail server. | {
"entities": {
"entity": [
{
"text": "Email",
"start": 0,
"end": 5
},
{
"text": "a specified mail server",
"start": 19,
"end": 42
}
],
"action": [
{
"text": "sent",
"start": 6,
"end": 10
}
],
... | [] |
One CHOPSTICK v1 variant contained modules and functions for collecting keystroke logs, Microsoft Office documents, and PGP files. | {
"entities": {
"entity": [
{
"text": "One CHOPSTICK v1 variant",
"start": 0,
"end": 24
},
{
"text": "keystroke logs",
"start": 72,
"end": 86
},
{
"text": "Microsoft Office documents",
"start": 88,
"end": 114
... | [] |
The monitoring for new files of interest is performed by a'' Directory Observer'' module. | {
"entities": {
"entity": [
{
"text": "The monitoring for new files of interest",
"start": 0,
"end": 40
},
{
"text": "a'' Directory Observer'' module",
"start": 57,
"end": 88
}
],
"action": [
{
"text": "is performed",
... | [] |
In one sample this information was intended to be sent via SMTP using a Georgian MIA mail server. | {
"entities": {
"entity": [
{
"text": "this information",
"start": 14,
"end": 30
},
{
"text": "SMTP",
"start": 59,
"end": 63
},
{
"text": "a Georgian MIA mail server",
"start": 70,
"end": 96
}
],
"a... | [] |
It used one of four embedded sender email addresses (@mia.gov.ge) to send files via email to another email address on the same mail server. | {
"entities": {
"entity": [
{
"text": "It",
"start": 0,
"end": 2
},
{
"text": "one of four embedded sender email addresses (@mia.gov.ge)",
"start": 8,
"end": 65
},
{
"text": "send files via email to another email address on the ... | [] |
All information required for the email was hardcoded in the backdoor. | {
"entities": {}
} | [] |
3. | {
"entities": {}
} | [] |
Local copying to defeat closed networks. | {
"entities": {}
} | [] |
One variant of CHOPSTICK focuses on apparent air gap/closed network capabilities by routing messages between local directories, the registry and USB drives. | {
"entities": {
"entity": [
{
"text": "One variant of CHOPSTICK",
"start": 0,
"end": 24
},
{
"text": "messages",
"start": 92,
"end": 100
},
{
"text": "local directories, the registry and USB drives",
"start": 109,
... | [] |
During our research into APT28's malware, we noted two details consistent across malware samples. | {
"entities": {}
} | [] |
The first was that APT28 had consistently compiled Russian language settings into their malware. | {
"entities": {}
} | [] |
The second was that malware compile times from 2007 to 2014 corresponded to normal business hours in the UTC + 4 time zone, which includes major Russian cities such as Moscow and St. Petersburg. | {
"entities": {}
} | [] |
PE resources include language information that can be helpful if a developer wants to show user interface items in a specific language. | {
"entities": {}
} | [] |
Non-default language settings packaged with PE resources are dependent on the developer's build environment. | {
"entities": {}
} | [] |
Each PE resource includes a'' locale'' identifier with a language ID'' composed of a primary language identifier indicating the language and a sublanguage identifier indicating the country/region. | {
"entities": {}
} | [] |
'' At the time of the writing of this paper, we had identified 103 malware samples that were both attributed to APT28 and contained PE resources. | {
"entities": {}
} | [] |
Table 5 shows the locale identifiers with associated language and country/region for these samples. | {
"entities": {}
} | [] |
The samples with Russian language settings were compiled between late 2007 and late 2013, as depicted in Figure 9. | {
"entities": {}
} | [] |
This consistency over a long timeframe suggests that the developers of APT28 malware were using a build environment with Russian language settings at least some of the time and made no effort to obscure this detail. | {
"entities": {}
} | [] |
Overall, the locale IDs suggest that APT28 developers can operate in both Russian and English. | {
"entities": {}
} | [] |
Of the 140 malware samples that we have attributed to APT28 so far, over 89% were compiled between 0400 and 1400 UTC time, as depicted in Figure 10. | {
"entities": {}
} | [] |
Over 96% were compiled between Monday and Friday. | {
"entities": {}
} | [] |
This parallels the working hours in UTC+0400 (that is, compile times begin about 8AM and end about 6PM in this time zone). | {
"entities": {}
} | [] |
This time zone includes major Russian cities such as Moscow and St. Petersburg. | {
"entities": {}
} | [] |
We started researching APT28 based on activity we observed on our clients' networks, similar to other targeted threat groups we have identified over time. | {
"entities": {}
} | [] |
We assess that APT28 is most likely sponsored by the Russian government. | {
"entities": {}
} | [] |
We summarize our key observations about APT28 in Figure 11 below. | {
"entities": {}
} | [] |
APT28's characteristics-their targeting, malware, language, and working hours-have led us to conclude that we are tracking a focused, long- standing espionage effort. | {
"entities": {}
} | [] |
Given the available data, we assess that APT28's work is sponsored by the Russian government. | {
"entities": {}
} | [] |
We use the term'' threat group'' to refer to actors who work together to target and penetrate networks of interest. | {
"entities": {}
} | [] |
These individuals may share the same set of tasks, coordinate targets, and share tools and methodology. | {
"entities": {}
} | [] |
They work together to gain access to their targets and steal data. | {
"entities": {
"entity": [
{
"text": "They",
"start": 0,
"end": 4
},
{
"text": "access to their targets",
"start": 27,
"end": 50
},
{
"text": "data",
"start": 61,
"end": 65
}
],
"action": [
{... | [] |
The art of attributing disparate intrusion activities to the same threat group is not always simple. | {
"entities": {}
} | [] |
Different groups may use similar intrusion methodologies and common tools, particularly those that are widely available on the Internet, such as pwdump, HTran, or Gh0st RAT. | {
"entities": {}
} | [] |
There may be overlaps between groups caused by the sharing of malware or exploits they have authored, or even the sharing of personnel. | {
"entities": {}
} | [] |
Individual threat actors may move between groups either temporarily or permanently. | {
"entities": {}
} | [] |
A threat actor may also be a private citizen who is hired by multiple groups. | {
"entities": {}
} | [] |
Multiple groups, on occasion, compromise the same target within the same timeframe. | {
"entities": {}
} | [] |
Distinguishing one threat group from another is possible with enough information, analytical experience, and tools to piece it all together. | {
"entities": {}
} | [] |
We can analyze multiple incidents and tell by the evidence left behind that a given incident was the result of one threat group and not another. | {
"entities": {}
} | [] |
Threat actors leave behind various forensic details. | {
"entities": {}
} | [] |
They may send spear phishing emails from a specific IP address or email address. | {
"entities": {
"entity": [
{
"text": "They",
"start": 0,
"end": 4
},
{
"text": "spear phishing emails",
"start": 14,
"end": 35
},
{
"text": "a specific IP address or email address",
"start": 41,
"end": 79
... | [] |
Their emails may contain certain patterns; files have specific names, MD5 hashes, timestamps, custom functions, and encryption algorithms. | {
"entities": {}
} | [] |
Their backdoors may have command and control IP addresses or domain names embedded. | {
"entities": {}
} | [] |
These are just a few examples of the myriad of forensic details that we consider when distinguishing one threat group from another. | {
"entities": {}
} | [] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.