mneb/malware-text-db · Datasets at Hugging Face

input stringlengths 1 785	output dict	schema listlengths 0 0
As reported earlier this year, Agent.BTZ used the same encoding key and the installation log file name as Uroburos.	{ "entities": {} }	[]
ComRAT, in its version 3.25, shows the same behavior.	{ "entities": {} }	[]
Furthermore, the attackers also shared a C & C domain.	{ "entities": { "entity": [ { "text": "the attackers", "start": 13, "end": 26 }, { "text": "a C & C domain", "start": 39, "end": 53 } ], "action": [ { "text": "shared", "start": 32, "end": 38 } ...	[]
The latest version of ComRAT known to us (v3.26) uses a new key and does not create the installation log file, in order to complicate the analysis and to disguise the link between the two cases.	{ "entities": { "entity": [ { "text": "ComRAT", "start": 22, "end": 28 }, { "text": "the analysis", "start": 134, "end": 146 } ], "action": [ { "text": "complicate", "start": 123, "end": 133 } ]...	[]
Another very interesting fact: the attackers use COM Object hijacking, the same persistence mechanism as COMpfun, which we described recently.	{ "entities": { "entity": [ { "text": "the attackers", "start": 31, "end": 44 }, { "text": "the same persistence mechanism", "start": 71, "end": 101 } ], "action": [ { "text": "use", "start": 45, "end": 4...	[]
Taken everything into consideration, the indications we saw during our analyzes lead to the supposition that the group behind Agent.BTZ and Uroburos is still active and is pursuing the Agent.BTZ path once more to improve and change the RAT.	{ "entities": {} }	[]
The analyzed file is the latest version we identified: v3.26.	{ "entities": {} }	[]
The version identification is described in the chapter'' Log files''.	{ "entities": {} }	[]
The major difference between this version and the older version (s) will be described there.	{ "entities": {} }	[]
The first task of the malware is to install the file credprov.tlb in% APPDATA% \Microsoft\.	{ "entities": { "entity": [ { "text": "the malware", "start": 18, "end": 29 }, { "text": "the file credprov.tlb", "start": 44, "end": 65 }, { "text": "% APPDATA% \\Microsoft\\", "start": 68, "end": 90 } ...	[]
This file is the main payload of the malware.	{ "entities": {} }	[]
The dropper executes the following command in order to install a second file: The second file is shdocw.tlp.	{ "entities": { "entity": [ { "text": "The dropper", "start": 0, "end": 11 }, { "text": "a second file", "start": 63, "end": 76 } ], "action": [ { "text": "install", "start": 55, "end": 62 } ] ...	[]
The two files are Microsoft Windows dynamic libraries.	{ "entities": {} }	[]
To be started during the boot process of the infected machine, the malware creates the following registry key: This registry key is used to associate the library shdocvw.tlp to the object 42aedc87-2188-41fd-b9a3-0c966feabec1 as previously explained in the article about COMpfun.	{ "entities": { "action": [ { "text": "started", "start": 6, "end": 13 }, { "text": "creates", "start": 75, "end": 82 } ], "modifier": [ { "text": "during", "start": 14, "end": 20 } ], "enti...	[]
The purpose is to load the library into each and every process executed on the infected system.	{ "entities": { "action": [ { "text": "load", "start": 18, "end": 22 } ], "entity": [ { "text": "the library", "start": 23, "end": 34 }, { "text": "each and every process executed on the infected system", "start"...	[]
If the version of the malware is older than 3.26, the dropper creates an additional file called winview.ocx.	{ "entities": { "entity": [ { "text": "the dropper", "start": 50, "end": 61 }, { "text": "an additional file called winview.ocx", "start": 70, "end": 107 } ], "action": [ { "text": "creates", "start": 62, ...	[]
We noticed that the file name is still the same as the file name used by Agent.BTZ in the past.	{ "entities": {} }	[]
The file is xored with the following obfuscation key (used by both, Uroburos and Agent.BTZ): Here is the decoded log file content: We can notice that the malware checks if an older version is installed on the system and if this is the case, the dropper removes the older version.	{ "entities": { "entity": [ { "text": "The file", "start": 0, "end": 8 }, { "text": "the following obfuscation key", "start": 23, "end": 52 }, { "text": "the malware", "start": 150, "end": 161 }, { ...	[]
In contrast to this, in our Uroburos analysis, we found out that Uroburos does not install itself in case a version of Agent.BTZ was found on a system.	{ "entities": { "entity": [ { "text": "Uroburos", "start": 65, "end": 73 }, { "text": "itself", "start": 91, "end": 97 }, { "text": "a version of Agent.BTZ", "start": 106, "end": 128 }, { "tex...	[]
During the startup of the infected machine, the shdocvw.tlp library is loaded into all processes.	{ "entities": { "entity": [ { "text": "the shdocvw.tlp library", "start": 44, "end": 67 }, { "text": "all processes", "start": 83, "end": 96 } ], "action": [ { "text": "is loaded", "start": 68, "end": 77 ...	[]
If the process is explorer.exe, this library will load the other library called credprov.tlb.	{ "entities": { "entity": [ { "text": "this library", "start": 32, "end": 44 }, { "text": "the other library called credprov.tlb", "start": 55, "end": 92 } ], "action": [ { "text": "will load", "start": 45, ...	[]
This library is the real payload.	{ "entities": {} }	[]
Its features are common for a Remote Administration Tool (RAT): ComRAT's communication to the command and control server is performed by the browser process and not by explorer.exe in order to avoid being blocked by a firewall on the system or any additional security products.	{ "entities": { "entity": [ { "text": "ComRAT's communication to the command and control server", "start": 64, "end": 120 }, { "text": "the browser process", "start": 137, "end": 156 }, { "text": "being blocked by a firewall on ...	[]
The communication between the processes is performed by named pipe.	{ "entities": { "entity": [ { "text": "The communication between the processes", "start": 0, "end": 39 }, { "text": "named pipe", "start": 56, "end": 66 } ], "action": [ { "text": "is performed", "start": 40, ...	[]
Two log files are created during the malware execution: mskfp32.ocx and msvcrtd.tlb.	{ "entities": { "entity": [ { "text": "Two log files", "start": 0, "end": 13 }, { "text": "the malware execution", "start": 33, "end": 54 }, { "text": "mskfp32.ocx and msvcrtd.tlb", "start": 56, "end": 83 }...	[]
If the malware version is older than 3.26, the xored key is the same as the dropper key.	{ "entities": {} }	[]
Concerning the version 3.26, the malware uses a new non-ASCII key.	{ "entities": {} }	[]
Here is an example of decoded log file for the version 3.26: We can identify the version of the malware thanks to the PVer flag.	{ "entities": {} }	[]
The command and control server information is stored in the registry, not in an XML, and encoded: For example, in the analyzed sample the CC is: weather-online.hopto.org.	{ "entities": { "entity": [ { "text": "The command and control server information", "start": 0, "end": 42 }, { "text": "the registry", "start": 56, "end": 68 } ], "action": [ { "text": "is stored", "start": 43, ...	[]
This domain is far from unknown, as it has been mentioned in BAE System's Uroburos (aka Snake) analysis paper as C & C server domain for the Uroburos malware.	{ "entities": {} }	[]
Another connection between the cases.	{ "entities": {} }	[]
If the malware version in lower than 3.26, the XML log file contains the command and control server information: Let us summarize the similarities and differences between Agent.BTZ, Uroburos and ComRAT as far as we can: Some parts of the code are exactly the same (appears to be copy & paste).	{ "entities": { "entity": [ { "text": "the XML log file", "start": 43, "end": 59 }, { "text": "the command and control server information", "start": 69, "end": 111 } ], "action": [ { "text": "contains", "start": ...	[]
That is the reason why the sample is detected as Uroburos (aka Turla).	{ "entities": {} }	[]
The same code was used by Agent.BTZ and also the dll loaded into userland during the Uroburos analysis.	{ "entities": {} }	[]
Command and control server domains are shared between Uroburos and ComRAT.	{ "entities": { "entity": [ { "text": "Command and control server domains", "start": 0, "end": 34 }, { "text": "Uroburos and ComRAT", "start": 54, "end": 73 } ], "action": [ { "text": "are shared", "start": 35, ...	[]
In version 3.26, the author changed the key and remove the known file name.	{ "entities": {} }	[]
This action can be an indication for the developer's effort to hide this connection.	{ "entities": {} }	[]
The main difference is the design.	{ "entities": {} }	[]
Agent.BTZ is a common RAT, a simple library executed on an infected machine.	{ "entities": { "entity": [ { "text": "Agent.BTZ", "start": 0, "end": 9 }, { "text": "a common RAT", "start": 13, "end": 25 } ], "action": [ { "text": "is", "start": 10, "end": 12 } ] } }	[]
ComRAT is more complex and cleverer.	{ "entities": {} }	[]
The malware is loaded into each and every process of the infected machine and the main part (payload) of the malware is only executed in explorer.exe.	{ "entities": { "entity": [ { "text": "The malware", "start": 0, "end": 11 }, { "text": "each and every process of the infected machine", "start": 27, "end": 73 }, { "text": "the main part (payload) of the malware", "sta...	[]
Furthermore, the C & C communication blends into the usual browser traffic and the malware communicates to the browser by named pipe.	{ "entities": { "entity": [ { "text": "the C & C communication", "start": 13, "end": 36 }, { "text": "the usual browser traffic", "start": 49, "end": 74 }, { "text": "the malware", "start": 79, "end": 90 },...	[]
It is by far a more complex userland design than Agent.BTZ.	{ "entities": {} }	[]
These differences, mainly the more complex design, lead us to give this malware a new name.	{ "entities": {} }	[]
The analyzed dropper of v3.25 has a compilation date of February 6th 2014.	{ "entities": {} }	[]
The more recent dropper of v3.26, which has all the mentioned changes implemented, reveals a compilation date of January 3rd 2013.	{ "entities": {} }	[]
We suspect that this date is spoofed in order to disguise that this is in fact a newer version.	{ "entities": { "entity": [ { "text": "this date", "start": 16, "end": 25 } ], "action": [ { "text": "is spoofed", "start": 26, "end": 36 } ] } }	[]
This analysis shows that even after the Uroburos publication in February 2014, the group behind this piece of malware seems to be still active.	{ "entities": {} }	[]
In any case, the ComRAT developers implemented new mechanisms, changed keys, removed log files to hide their activities and tried to disguise the connections between the RAT ComRAT, the rootkit Uroburos and the RAT Agent.BTZ as much as possible.	{ "entities": {} }	[]
However, we can still follow the evolution of the malware by comparing the versions.	{ "entities": {} }	[]
The persistence mechanism discovered in October 2014 makes it possible to intrude into a system in a really discreet manner and we estimate that other actors will use the same persistence mechanism in the near future.	{ "entities": {} }	[]
We will definitely keep our ears and eyes open and continue analyzing.	{ "entities": {} }	[]
Repressive nation-states have long monitored telecommunica- tions to keep tabs on political dissent.	{ "entities": {} }	[]
The Internet and online social networks, however, pose novel technical challenges to this practice, even as they open up new domains for surveil- lance.	{ "entities": {} }	[]
We analyze an extensive collection of suspicious files and links targeting activists, opposition members, and non- governmental organizations in the Middle East over the past several years.	{ "entities": {} }	[]
We find that these artifacts reflect efforts to at- tack targets' devices for the purposes of eavesdropping, stealing information, and/or unmasking anonymous users.	{ "entities": {} }	[]
We describe attack campaigns we have observed in Bahrain, Syria, and the United Arab Emirates, investigating attackers, tools, and tech- niques.	{ "entities": {} }	[]
In addition to off-the-shelf remote access trojans and the use of third-party IP-tracking services, we identify commer- cial spyware marketed exclusively to governments, including Gamma's FinSpy and Hacking Team's Remote Control Sys- tem (RCS).	{ "entities": {} }	[]
We describe their use in Bahrain and the UAE, and map out the potential broader scope of this activity by conduct- ing global scans of the corresponding command-and-control (C & C) servers.	{ "entities": {} }	[]
Finally, we frame the real-world consequences of these campaigns via strong circumstantial evidence linking hacking to arrests, interrogations, and imprisonment.	{ "entities": {} }	[]
Computer security research devotes extensive efforts to pro- tecting individuals against indiscriminate, large-scale attacks such as those used by cybercriminals.	{ "entities": {} }	[]
Recently, the prob- lem of protecting institutions against targeted attacks conducted by nation-states (so-called'' Advanced Persistent Threats'') has likewise elicited significant research interest.	{ "entities": {} }	[]
Where these two problem domains intersect, however-targeted cyber attacks by nation-states against individuals-has received virtually no sig- nificant, methodical research attention to date.	{ "entities": {} }	[]
This new prob- lem space poses challenges that are both technically complex and of significant real-world importance.	{ "entities": {} }	[]
In this work we undertake to characterize the emergent prob- lem space of nation-state Internet attacks against individuals engaged in pro-democracy or opposition movements.	{ "entities": {} }	[]
While we lack the data to do so in a fully comprehensive fashion, we provide extensive detail from both technical and operational perspectives as seen in three countries.	{ "entities": {} }	[]
We view such character- izations as the fundamental first step necessary for the rigorous, scientific pursuit of a new problem space.	{ "entities": {} }	[]
For our study we draw upon several years of research we have conducted into cases from Bahrain, Syria and the United Arab Emirates.	{ "entities": {} }	[]
We frame the nature of these attacks, and the technology and infrastructure used to conduct them, in the con- text of their impacts on real people.	{ "entities": {} }	[]
We hope in the process to inspire additional research efforts addressing the difficult prob- lem of how to adequately protect individuals with very limited resources facing powerful adversaries.	{ "entities": {} }	[]
As an illustration of this phenomenon, consider the follow- ing anecdote, pieced together from public reports and court documents.	{ "entities": {} }	[]
At dawn on 3/12/13, police raided the house of 17-year- old Ali Al-Shofa, confiscated his laptop and phone, and took him into custody.	{ "entities": {} }	[]
He was charged with referring to Bahrain's King as a'' dictator'' () and'' fallen one'' () on a pseudonymous Twitter account, @alkawarahnews.	{ "entities": {} }	[]
Accord- ing to court documents, Bahrain's Cyber Crime Unit had linked an IP address registered in his father's name to the account on 12/9/12.	{ "entities": {} }	[]
Operators of @alkawarahnews later forwarded a suspicious private message to one of the authors.	{ "entities": {} }	[]
The message was received on 12/8/12 on a Facebook account linked to the Twitter handle, and contained a link to a protest video, purport- edly sent by an anti-government individual.	{ "entities": {} }	[]
The link redirected through iplogger.org, a service that records the IP address of anyone who clicks.	{ "entities": {} }	[]
Analytics for the link indicate that it had been clicked once from inside Bahrain.	{ "entities": {} }	[]
On 6/25/13, Ali was sentenced to one year in prison.	{ "entities": {} }	[]
Ali's case is an example of the larger phenomenon we in- vestigate: attacks against activists, dissidents, trade unionists, human rights campaigners, journalists, and members of NGOs (henceforth'' targets'') in the Middle East.	{ "entities": {} }	[]
The attacks we have documented usually involve the use of malicious links or e-mail attachments, designed to obtain information from a device.	{ "entities": { "entity": [ { "text": "The attacks we have documented", "start": 0, "end": 30 }, { "text": "the use of malicious links or e-mail attachments", "start": 47, "end": 95 }, { "text": "information", "start": 1...	[]
On the one hand, we have observed attacks using a wide range of off-the-shelf spyware, as well as publicly available third-party services, like iplogger.org.	{ "entities": { "entity": [ { "text": "attacks", "start": 34, "end": 41 }, { "text": "a wide range of off-the-shelf spyware, as well as publicly available third-party services, like iplogger.org", "start": 48, "end": 156 } ], "action"...	[]
On the other hand, some at- tacks use so-called'' lawful intercept'' trojans and related equip- ment, purportedly sold exclusively to governments by compa- nies like Gamma International and Hacking Team.	{ "entities": { "entity": [ { "text": "some at- tacks", "start": 19, "end": 33 }, { "text": "so-called'' lawful intercept'' trojans", "start": 38, "end": 76 } ], "action": [ { "text": "use", "start": 34, ...	[]
The lat- ter advertises that governments need its technology to'' look through their target's eyes'' rather than rely solely on'' passive monitoring''.	{ "entities": {} }	[]
Overall, the attacks we document are rarely technically novel.	{ "entities": {} }	[]
In fact, we suspect that the majority of at- tacks could be substantially limited via well-known security practices, settings, and software updates.	{ "entities": {} }	[]
Yet, the attacks are noteworthy for their careful social engineering, their links to governments, and their real-world impact.	{ "entities": { "entity": [ { "text": "the attacks", "start": 5, "end": 16 } ], "action": [ { "text": "careful social engineering", "start": 42, "end": 68 } ] } }	[]
We obtained the majority of our artifacts by encouraging in- dividuals who might be targeted by governments to provide us with suspicious files and unsolicited links, especially from un- familiar senders.	{ "entities": {} }	[]
While this process has provided a rich set of artifacts to analyze, it does not permit us to claim our dataset is representative.	{ "entities": {} }	[]
Our analysis links these attacks with a common class of ac- tor: an attacker whose behavior, choice of target, or use of in- formation obtained in the attack, aligns with the interests of a government.	{ "entities": {} }	[]
In some cases, such as Ali's, the attackers appear to be governments themselves; in other cases, they appear in- stead to be pro-government actors, ranging from patriotic, not necessarily skilled volunteers to cyber mercenaries.	{ "entities": {} }	[]
The phe- nomenon has been identified before, such as in Libya, when the fall of Gaddafi's regime revealed direct government ties to hacking during the 2011 Civil War.	{ "entities": {} }	[]
We make the following contributions: • We analyze the technology associated with targeted at- tacks (e.g., malicious links, spyware), and trace it back to its programmers and manufacturers.	{ "entities": {} }	[]
While the attacks are not novel-and indeed often involve technology used by the cybercrime underground-they are significant be- cause they have a real-world impact and visibility, and are connected to governments.	{ "entities": {} }	[]
In addition, we often find amateurish mistakes in either the attacker's technology or operations, indicating that energy spent countering these threats can realize significant benefits.	{ "entities": {} }	[]
We do not, how- ever, conclude that all nation-state attacks or attackers are incompetent, and we suspect that some attacks have evaded our detection.	{ "entities": {} }	[]
• When possible, we empirically characterize the attacks and technology we have observed.	{ "entities": {} }	[]
We map out global use of two commercial hacking tools by governments by searching through Internet scan data using fingerprints for command-and-control (C & C) servers derived from our spyware analysis.	{ "entities": {} }	[]
• We develop strong evidence tying attacks to govern- ment sponsors and corporate suppliers, countering de- nials, sometimes energetic and sometimes indirect, of such involvement [,,,], in contrast to denials or claims of a corporate'' oversight'' board.	{ "entities": {} }	[]
Our scan- ning suggests use of'' lawful intercept'' trojans by 11 ad- ditional countries considered governed by'' authoritarian regimes.	{ "entities": {} }	[]

As reported earlier this year, Agent.BTZ used the same encoding key and the installation log file name as Uroburos.

{ "entities": {} }

[]

ComRAT, in its version 3.25, shows the same behavior.

{ "entities": {} }

[]

Furthermore, the attackers also shared a C & C domain.

{ "entities": { "entity": [ { "text": "the attackers", "start": 13, "end": 26 }, { "text": "a C & C domain", "start": 39, "end": 53 } ], "action": [ { "text": "shared", "start": 32, "end": 38 } ...

[]

The latest version of ComRAT known to us (v3.26) uses a new key and does not create the installation log file, in order to complicate the analysis and to disguise the link between the two cases.

{ "entities": { "entity": [ { "text": "ComRAT", "start": 22, "end": 28 }, { "text": "the analysis", "start": 134, "end": 146 } ], "action": [ { "text": "complicate", "start": 123, "end": 133 } ]...

[]

Another very interesting fact: the attackers use COM Object hijacking, the same persistence mechanism as COMpfun, which we described recently.

{ "entities": { "entity": [ { "text": "the attackers", "start": 31, "end": 44 }, { "text": "the same persistence mechanism", "start": 71, "end": 101 } ], "action": [ { "text": "use", "start": 45, "end": 4...

[]

Taken everything into consideration, the indications we saw during our analyzes lead to the supposition that the group behind Agent.BTZ and Uroburos is still active and is pursuing the Agent.BTZ path once more to improve and change the RAT.

{ "entities": {} }

[]

The analyzed file is the latest version we identified: v3.26.

{ "entities": {} }

[]

The version identification is described in the chapter'' Log files''.

{ "entities": {} }

[]

The major difference between this version and the older version (s) will be described there.

{ "entities": {} }

[]

The first task of the malware is to install the file credprov.tlb in% APPDATA% \Microsoft\.

{ "entities": { "entity": [ { "text": "the malware", "start": 18, "end": 29 }, { "text": "the file credprov.tlb", "start": 44, "end": 65 }, { "text": "% APPDATA% \\Microsoft\\", "start": 68, "end": 90 } ...

[]

This file is the main payload of the malware.

{ "entities": {} }

[]

The dropper executes the following command in order to install a second file: The second file is shdocw.tlp.

{ "entities": { "entity": [ { "text": "The dropper", "start": 0, "end": 11 }, { "text": "a second file", "start": 63, "end": 76 } ], "action": [ { "text": "install", "start": 55, "end": 62 } ] ...

[]

The two files are Microsoft Windows dynamic libraries.

{ "entities": {} }

[]

To be started during the boot process of the infected machine, the malware creates the following registry key: This registry key is used to associate the library shdocvw.tlp to the object 42aedc87-2188-41fd-b9a3-0c966feabec1 as previously explained in the article about COMpfun.

{ "entities": { "action": [ { "text": "started", "start": 6, "end": 13 }, { "text": "creates", "start": 75, "end": 82 } ], "modifier": [ { "text": "during", "start": 14, "end": 20 } ], "enti...

[]

The purpose is to load the library into each and every process executed on the infected system.

{ "entities": { "action": [ { "text": "load", "start": 18, "end": 22 } ], "entity": [ { "text": "the library", "start": 23, "end": 34 }, { "text": "each and every process executed on the infected system", "start"...

[]

If the version of the malware is older than 3.26, the dropper creates an additional file called winview.ocx.

{ "entities": { "entity": [ { "text": "the dropper", "start": 50, "end": 61 }, { "text": "an additional file called winview.ocx", "start": 70, "end": 107 } ], "action": [ { "text": "creates", "start": 62, ...

[]

We noticed that the file name is still the same as the file name used by Agent.BTZ in the past.

{ "entities": {} }

[]

The file is xored with the following obfuscation key (used by both, Uroburos and Agent.BTZ): Here is the decoded log file content: We can notice that the malware checks if an older version is installed on the system and if this is the case, the dropper removes the older version.

{ "entities": { "entity": [ { "text": "The file", "start": 0, "end": 8 }, { "text": "the following obfuscation key", "start": 23, "end": 52 }, { "text": "the malware", "start": 150, "end": 161 }, { ...

[]

In contrast to this, in our Uroburos analysis, we found out that Uroburos does not install itself in case a version of Agent.BTZ was found on a system.

{ "entities": { "entity": [ { "text": "Uroburos", "start": 65, "end": 73 }, { "text": "itself", "start": 91, "end": 97 }, { "text": "a version of Agent.BTZ", "start": 106, "end": 128 }, { "tex...

[]

During the startup of the infected machine, the shdocvw.tlp library is loaded into all processes.

{ "entities": { "entity": [ { "text": "the shdocvw.tlp library", "start": 44, "end": 67 }, { "text": "all processes", "start": 83, "end": 96 } ], "action": [ { "text": "is loaded", "start": 68, "end": 77 ...

[]

If the process is explorer.exe, this library will load the other library called credprov.tlb.

{ "entities": { "entity": [ { "text": "this library", "start": 32, "end": 44 }, { "text": "the other library called credprov.tlb", "start": 55, "end": 92 } ], "action": [ { "text": "will load", "start": 45, ...

[]

This library is the real payload.

{ "entities": {} }

[]

Its features are common for a Remote Administration Tool (RAT): ComRAT's communication to the command and control server is performed by the browser process and not by explorer.exe in order to avoid being blocked by a firewall on the system or any additional security products.

{ "entities": { "entity": [ { "text": "ComRAT's communication to the command and control server", "start": 64, "end": 120 }, { "text": "the browser process", "start": 137, "end": 156 }, { "text": "being blocked by a firewall on ...

[]

The communication between the processes is performed by named pipe.

{ "entities": { "entity": [ { "text": "The communication between the processes", "start": 0, "end": 39 }, { "text": "named pipe", "start": 56, "end": 66 } ], "action": [ { "text": "is performed", "start": 40, ...

[]

Two log files are created during the malware execution: mskfp32.ocx and msvcrtd.tlb.

{ "entities": { "entity": [ { "text": "Two log files", "start": 0, "end": 13 }, { "text": "the malware execution", "start": 33, "end": 54 }, { "text": "mskfp32.ocx and msvcrtd.tlb", "start": 56, "end": 83 }...

[]

If the malware version is older than 3.26, the xored key is the same as the dropper key.

{ "entities": {} }

[]

Concerning the version 3.26, the malware uses a new non-ASCII key.

{ "entities": {} }

[]

Here is an example of decoded log file for the version 3.26: We can identify the version of the malware thanks to the PVer flag.

{ "entities": {} }

[]

The command and control server information is stored in the registry, not in an XML, and encoded: For example, in the analyzed sample the CC is: weather-online.hopto.org.

{ "entities": { "entity": [ { "text": "The command and control server information", "start": 0, "end": 42 }, { "text": "the registry", "start": 56, "end": 68 } ], "action": [ { "text": "is stored", "start": 43, ...

[]

This domain is far from unknown, as it has been mentioned in BAE System's Uroburos (aka Snake) analysis paper as C & C server domain for the Uroburos malware.

{ "entities": {} }

[]

Another connection between the cases.

{ "entities": {} }

[]

If the malware version in lower than 3.26, the XML log file contains the command and control server information: Let us summarize the similarities and differences between Agent.BTZ, Uroburos and ComRAT as far as we can: Some parts of the code are exactly the same (appears to be copy & paste).

{ "entities": { "entity": [ { "text": "the XML log file", "start": 43, "end": 59 }, { "text": "the command and control server information", "start": 69, "end": 111 } ], "action": [ { "text": "contains", "start": ...

[]

That is the reason why the sample is detected as Uroburos (aka Turla).

{ "entities": {} }

[]

The same code was used by Agent.BTZ and also the dll loaded into userland during the Uroburos analysis.

{ "entities": {} }

[]

Command and control server domains are shared between Uroburos and ComRAT.

{ "entities": { "entity": [ { "text": "Command and control server domains", "start": 0, "end": 34 }, { "text": "Uroburos and ComRAT", "start": 54, "end": 73 } ], "action": [ { "text": "are shared", "start": 35, ...

[]

In version 3.26, the author changed the key and remove the known file name.

{ "entities": {} }

[]

This action can be an indication for the developer's effort to hide this connection.

{ "entities": {} }

[]

The main difference is the design.

{ "entities": {} }

[]

Agent.BTZ is a common RAT, a simple library executed on an infected machine.

{ "entities": { "entity": [ { "text": "Agent.BTZ", "start": 0, "end": 9 }, { "text": "a common RAT", "start": 13, "end": 25 } ], "action": [ { "text": "is", "start": 10, "end": 12 } ] } }

[]

ComRAT is more complex and cleverer.

{ "entities": {} }

[]

The malware is loaded into each and every process of the infected machine and the main part (payload) of the malware is only executed in explorer.exe.

{ "entities": { "entity": [ { "text": "The malware", "start": 0, "end": 11 }, { "text": "each and every process of the infected machine", "start": 27, "end": 73 }, { "text": "the main part (payload) of the malware", "sta...

[]

Furthermore, the C & C communication blends into the usual browser traffic and the malware communicates to the browser by named pipe.

{ "entities": { "entity": [ { "text": "the C & C communication", "start": 13, "end": 36 }, { "text": "the usual browser traffic", "start": 49, "end": 74 }, { "text": "the malware", "start": 79, "end": 90 },...

[]

It is by far a more complex userland design than Agent.BTZ.

{ "entities": {} }

[]

These differences, mainly the more complex design, lead us to give this malware a new name.

{ "entities": {} }

[]

The analyzed dropper of v3.25 has a compilation date of February 6th 2014.

{ "entities": {} }

[]

The more recent dropper of v3.26, which has all the mentioned changes implemented, reveals a compilation date of January 3rd 2013.

{ "entities": {} }

[]

We suspect that this date is spoofed in order to disguise that this is in fact a newer version.

{ "entities": { "entity": [ { "text": "this date", "start": 16, "end": 25 } ], "action": [ { "text": "is spoofed", "start": 26, "end": 36 } ] } }

[]

This analysis shows that even after the Uroburos publication in February 2014, the group behind this piece of malware seems to be still active.

{ "entities": {} }

[]

In any case, the ComRAT developers implemented new mechanisms, changed keys, removed log files to hide their activities and tried to disguise the connections between the RAT ComRAT, the rootkit Uroburos and the RAT Agent.BTZ as much as possible.

{ "entities": {} }

[]

However, we can still follow the evolution of the malware by comparing the versions.

{ "entities": {} }

[]

The persistence mechanism discovered in October 2014 makes it possible to intrude into a system in a really discreet manner and we estimate that other actors will use the same persistence mechanism in the near future.

{ "entities": {} }

[]

We will definitely keep our ears and eyes open and continue analyzing.

{ "entities": {} }

[]

Repressive nation-states have long monitored telecommunica- tions to keep tabs on political dissent.

{ "entities": {} }

[]

The Internet and online social networks, however, pose novel technical challenges to this practice, even as they open up new domains for surveil- lance.

{ "entities": {} }

[]

We analyze an extensive collection of suspicious files and links targeting activists, opposition members, and non- governmental organizations in the Middle East over the past several years.

{ "entities": {} }

[]

We find that these artifacts reflect efforts to at- tack targets' devices for the purposes of eavesdropping, stealing information, and/or unmasking anonymous users.

{ "entities": {} }

[]

We describe attack campaigns we have observed in Bahrain, Syria, and the United Arab Emirates, investigating attackers, tools, and tech- niques.

{ "entities": {} }

[]

In addition to off-the-shelf remote access trojans and the use of third-party IP-tracking services, we identify commer- cial spyware marketed exclusively to governments, including Gamma's FinSpy and Hacking Team's Remote Control Sys- tem (RCS).

{ "entities": {} }

[]

We describe their use in Bahrain and the UAE, and map out the potential broader scope of this activity by conduct- ing global scans of the corresponding command-and-control (C & C) servers.

{ "entities": {} }

[]

Finally, we frame the real-world consequences of these campaigns via strong circumstantial evidence linking hacking to arrests, interrogations, and imprisonment.

{ "entities": {} }

[]

Computer security research devotes extensive efforts to pro- tecting individuals against indiscriminate, large-scale attacks such as those used by cybercriminals.

{ "entities": {} }

[]

Recently, the prob- lem of protecting institutions against targeted attacks conducted by nation-states (so-called'' Advanced Persistent Threats'') has likewise elicited significant research interest.

{ "entities": {} }

[]

Where these two problem domains intersect, however-targeted cyber attacks by nation-states against individuals-has received virtually no sig- nificant, methodical research attention to date.

{ "entities": {} }

[]

This new prob- lem space poses challenges that are both technically complex and of significant real-world importance.

{ "entities": {} }

[]

In this work we undertake to characterize the emergent prob- lem space of nation-state Internet attacks against individuals engaged in pro-democracy or opposition movements.

{ "entities": {} }

[]

While we lack the data to do so in a fully comprehensive fashion, we provide extensive detail from both technical and operational perspectives as seen in three countries.

{ "entities": {} }

[]

We view such character- izations as the fundamental first step necessary for the rigorous, scientific pursuit of a new problem space.

{ "entities": {} }

[]

For our study we draw upon several years of research we have conducted into cases from Bahrain, Syria and the United Arab Emirates.

{ "entities": {} }

[]

We frame the nature of these attacks, and the technology and infrastructure used to conduct them, in the con- text of their impacts on real people.

{ "entities": {} }

[]

We hope in the process to inspire additional research efforts addressing the difficult prob- lem of how to adequately protect individuals with very limited resources facing powerful adversaries.

{ "entities": {} }

[]

As an illustration of this phenomenon, consider the follow- ing anecdote, pieced together from public reports and court documents.

{ "entities": {} }

[]

At dawn on 3/12/13, police raided the house of 17-year- old Ali Al-Shofa, confiscated his laptop and phone, and took him into custody.

{ "entities": {} }

[]

He was charged with referring to Bahrain's King as a'' dictator'' () and'' fallen one'' () on a pseudonymous Twitter account, @alkawarahnews.

{ "entities": {} }

[]

Accord- ing to court documents, Bahrain's Cyber Crime Unit had linked an IP address registered in his father's name to the account on 12/9/12.

{ "entities": {} }

[]

Operators of @alkawarahnews later forwarded a suspicious private message to one of the authors.

{ "entities": {} }

[]

The message was received on 12/8/12 on a Facebook account linked to the Twitter handle, and contained a link to a protest video, purport- edly sent by an anti-government individual.

{ "entities": {} }

[]

The link redirected through iplogger.org, a service that records the IP address of anyone who clicks.

{ "entities": {} }

[]

Analytics for the link indicate that it had been clicked once from inside Bahrain.

{ "entities": {} }

[]

On 6/25/13, Ali was sentenced to one year in prison.

{ "entities": {} }

[]

Ali's case is an example of the larger phenomenon we in- vestigate: attacks against activists, dissidents, trade unionists, human rights campaigners, journalists, and members of NGOs (henceforth'' targets'') in the Middle East.

{ "entities": {} }

[]

The attacks we have documented usually involve the use of malicious links or e-mail attachments, designed to obtain information from a device.

{ "entities": { "entity": [ { "text": "The attacks we have documented", "start": 0, "end": 30 }, { "text": "the use of malicious links or e-mail attachments", "start": 47, "end": 95 }, { "text": "information", "start": 1...

[]

On the one hand, we have observed attacks using a wide range of off-the-shelf spyware, as well as publicly available third-party services, like iplogger.org.

{ "entities": { "entity": [ { "text": "attacks", "start": 34, "end": 41 }, { "text": "a wide range of off-the-shelf spyware, as well as publicly available third-party services, like iplogger.org", "start": 48, "end": 156 } ], "action"...

[]

On the other hand, some at- tacks use so-called'' lawful intercept'' trojans and related equip- ment, purportedly sold exclusively to governments by compa- nies like Gamma International and Hacking Team.

{ "entities": { "entity": [ { "text": "some at- tacks", "start": 19, "end": 33 }, { "text": "so-called'' lawful intercept'' trojans", "start": 38, "end": 76 } ], "action": [ { "text": "use", "start": 34, ...

[]

The lat- ter advertises that governments need its technology to'' look through their target's eyes'' rather than rely solely on'' passive monitoring''.

{ "entities": {} }

[]

Overall, the attacks we document are rarely technically novel.

{ "entities": {} }

[]

In fact, we suspect that the majority of at- tacks could be substantially limited via well-known security practices, settings, and software updates.

{ "entities": {} }

[]

Yet, the attacks are noteworthy for their careful social engineering, their links to governments, and their real-world impact.

{ "entities": { "entity": [ { "text": "the attacks", "start": 5, "end": 16 } ], "action": [ { "text": "careful social engineering", "start": 42, "end": 68 } ] } }

[]

We obtained the majority of our artifacts by encouraging in- dividuals who might be targeted by governments to provide us with suspicious files and unsolicited links, especially from un- familiar senders.

{ "entities": {} }

[]

While this process has provided a rich set of artifacts to analyze, it does not permit us to claim our dataset is representative.

{ "entities": {} }

[]

Our analysis links these attacks with a common class of ac- tor: an attacker whose behavior, choice of target, or use of in- formation obtained in the attack, aligns with the interests of a government.

{ "entities": {} }

[]

In some cases, such as Ali's, the attackers appear to be governments themselves; in other cases, they appear in- stead to be pro-government actors, ranging from patriotic, not necessarily skilled volunteers to cyber mercenaries.

{ "entities": {} }

[]

The phe- nomenon has been identified before, such as in Libya, when the fall of Gaddafi's regime revealed direct government ties to hacking during the 2011 Civil War.

{ "entities": {} }

[]

We make the following contributions: • We analyze the technology associated with targeted at- tacks (e.g., malicious links, spyware), and trace it back to its programmers and manufacturers.

{ "entities": {} }

[]

While the attacks are not novel-and indeed often involve technology used by the cybercrime underground-they are significant be- cause they have a real-world impact and visibility, and are connected to governments.

{ "entities": {} }

[]

In addition, we often find amateurish mistakes in either the attacker's technology or operations, indicating that energy spent countering these threats can realize significant benefits.

{ "entities": {} }

[]

We do not, how- ever, conclude that all nation-state attacks or attackers are incompetent, and we suspect that some attacks have evaded our detection.

{ "entities": {} }

[]

• When possible, we empirically characterize the attacks and technology we have observed.

{ "entities": {} }

[]

We map out global use of two commercial hacking tools by governments by searching through Internet scan data using fingerprints for command-and-control (C & C) servers derived from our spyware analysis.

{ "entities": {} }

[]

• We develop strong evidence tying attacks to govern- ment sponsors and corporate suppliers, countering de- nials, sometimes energetic and sometimes indirect, of such involvement [,,,], in contrast to denials or claims of a corporate'' oversight'' board.

{ "entities": {} }

[]

Our scan- ning suggests use of'' lawful intercept'' trojans by 11 ad- ditional countries considered governed by'' authoritarian regimes.

{ "entities": {} }

[]