July 24, 2019 • Insikt Group®
Click here to download the complete analysis as a PDF.
Recorded Future’s Insikt GroupⓇ analyzed advertisements and comments within underground forums to determine popular malware and malware categories within underground forums. Sources include the Recorded FutureⓇ Platform, as well as open web, dark web, and underground forum research.
This report will be of greatest interest to organizations seeking to better understand malware dissemination within the criminal underground, as well as those who wish to monitor developing malware-related criminal threats.
By analyzing over 3.9 million posts from May 2018 to May 2019 across all underground forums indexed by the Recorded Future platform, Insikt Group identified the top malware variants being referenced on underground forums. Insikt Group also attempted to find real-world events that correlated with a higher number of malware references on these forums, as well as differences in tools advertised in forums of different languages, to see if any differences existed.
Insikt Group discovered that a majority of the top 10 mentions of malware in multiple languages included openly available dual-use tools, open-source malware, or cracked malware. Some of these malware families were also over three years old or could be mitigated with basic security precautions. Activity in underground forums that correlated to growth in malware references included: sale of malware in a larger bundle, advertising updates to the malware, advertisements of the malware on a new forum in which the malware was not previously sold, news articles related to malware shared on forums, and community engagement.
Insikt Group also discovered that underground communities in different languages did indeed focus on different malware, malware categories, and attack vectors. English- and Chinese-speaking underground communities, for example, focused more on Android malware than other communities. By separating forum advertisements by language, Insikt Group found that forum members occasionally used online translation services to attract business partners and buyers from different language communities.
In the last year, Recorded Future has reported on the hacking communities within Russia, China, Iran, and Brazil. This research draws special attention to what malware is popular within those communities. In order to better understand commodity malware that may be targeting client environments, we used a data-driven approach to answer the following questions:
From May 1, 2018, to May 1, 2019, Insikt Group researchers pulled all mentions of any malware family or category by month from underground forums spanning the dark web, the open web, and related sources. Insikt Group gathered over 3.9 million posts and checked the posts for mentions of 61 malware categories and 101,124 malware names.
The definition of malware used in this report is “operational pieces of code used to conduct illegal activity.” This expanded the scope of the data set to include vague categories (e.g., botnet, crypter, or webshell), red-teaming tools, or even certain exploits, such as ETERNALBLUE. While these entities are not usually classified as malware, individuals in both underground markets and forums talk about these entities similarly enough to malware to be notable, as shown by the data in the “Threat Analysis” section below.
There were four limitations inherent within the data set:
After gathering the data, we separated the number of mentions over the 13 months by eight languages, and turned the top 10 mentions of malware by language into bar graphs (for the full set of graphs, see Appendix B).
Overall, we observed that a majority of the top 10 graphs included openly available dual-use tools, such as MinerGate and Imminent Monitor (a cryptominer and a remote-access tool, respectively, created initially for legitimate uses), and open-source malware, such as njRat, AhMyth, and Mirai. This likely demonstrates that underground forum members are eager to discuss and use tools readily available to them rather than pay for or invent new tools. Many of the non open-sourced entities mentioned, like SpyNote, Trillium, NLBrute, and RDPBrute, had been previously cracked, meaning that multiple forum members now distribute unauthorized copies of the malware, usually at cheaper prices than the original seller.
The top 10 graphs also included malware that had been around for over three years, like Gh0st RAT, in addition to malware that is usually detectable with antivirus software or thwarted with good password hygiene. For example, RDPBrute (and its variants) will brute-force usernames and passwords on IPs with open RDP ports to gain initial access on a machine. This tool could be easily thwarted with difficult passwords, or by turning off RDP entirely. However, forum members continue to use this tool (and others) regardless, suggesting that they have been able to successfully infect victim hosts with the above malware.
Additionally, the graphs included tools designed to conduct other illegal activity, such as account stuffing, spamming, or carding. For example, Xrumer is software that allows criminals to spam multiple forums and forum comments with similar posts in an attempt to improve their results on forum search engines, or even regular search engines on the open web.
We discovered several pieces of malware that were discussed broadly within multiple language groups, including:
While many of the malware presented in our analysis had similar characteristics, each bar graph contained remarkably different content, showing that underground forums of different languages focus on different families of malware. Mentions of Nanocore, a cheap and easy-to-use remote-access trojan, were discussed more frequently on Farsi and Japanese forums. Xrumer had a high number of references on Russian forums and far fewer references on English forums, but was found in no other languages. The top malware entity mentioned in English-speaking forums was Trillium Security Multisploit Tool, which has only shown up a handful of times in forums containing other languages.
Additionally, the data shows that underground forums of different languages focus on different targets and attack vectors. For example, Chinese- and English-speaking underground forums focus more on targeting Android devices than their Russian counterparts. The top 10 malware within the Chinese-speaking underground included three Android trojans: SpyNote, AhMyth, and DroidJack. The English-speaking underground included two of those three: SpyNote and DroidJack. This is in stark contrast to the Russian-language group, whose top 10 contains no mobile malware whatsoever.
As mentioned in the research limitations outlined above, certain language-based data was inherently skewed, as individuals posting on forums would post in their non-native language if the forum operated primarily in a certain language (for example, a Russian forum member posting in English on an English-language forum), or if the individual wanted to attract buyers from a specific country. The latter discovery adds an additional layer to our previous research comparing Chinese and Russian underground forums, in which Recorded Future found that Russian- and English-language forums also contained Chinese-language posts, and assessed that Chinese forum members were likely advertising services to other Chinese individuals on non-Chinese forums. While certain posts on non-Chinese forums were written in native Chinese, other forum posts were likely non-Chinese individuals using Google Translate to attract Chinese- or English-speaking business partners and buyers.
While mentions of malware is useful data to analyze in some regard, raw mentions do not count comments on a forum post about the family of malware, which usually represent other individuals that are interested in buying or discussing the malware.
To address this issue, Insikt Group measured malware popularity using the following metrics:
The complete equation for the popularity of a single family of malware across X number of underground forums is as follows:
Effectively, for every forum, the sum of every mention and thread reply related to the malware in a single forum would be calculated. The sums for each forum would then be added together to get the total popularity.
By calculating metrics using the equation with the data previously gathered, Insikt Group was able to obtain the results shown below.
From May 2018 to May 2019, the top malware category mentioned in underground forums was ransomware, followed by crypter, and finally, trojan (with approximately two-thirds as many mentions). While “botnet” is not a malware category in a technical sense, botnets had more than twice as many mentions as ransomware. Based on the data, Insikt Group assesses that a large portion of individuals on underground forums likely still collect and use botnets for a variety of illicit activity, including DDoS attacks, cryptomining, fraud, and sending spam.
Of the top 10 malware mentions over all indexed forums, five (njRAT, SpyNote, DarkComet, Imminent Monitor, and WARZONE RAT) are remote-access trojans, two (Predator the Thief and AZORult) are information stealers, one is a tool used for initial access to a victim (NLBrute), and one (Xrumer) is a forum-specific tool.
Additionally, while “ransomware” was the top malware category mentioned on underground forums in the last year, it is worth noting that only one of the top 10 specific malware strains mentioned, GandCrab, is a ransomware strain. Out of the top 150 strains of malware collected, only 11 were ransomware, as shown in the table below. Approximately 50% of the mentions on underground forums in the past year were discussions and sales posts on generic, lower-level ransomware that do not have names or branding. This supports a previous analysis that Recorded Future published on this topic, looking toward 2019 ransomware trends.
|6: GandCrab||30: WannaCry||51: Cryptolocker||73: Princess Locker|
|84: Petya||88: NotPetya||110: Relock||124: Saturn Ransomware|
|125: Samsam||133: Ryuk Ransomware||144: Rapid Ransomware|
The graph below illustrates the same top 10 strains of malware on underground forums, but separates the number of posts by month. As shown in the graph, our analysis indicates that the families of malware that were discussed most frequently over the course of the year had different levels of engagement over the 13-month period. After taking a deeper dive into the content of the forum threads during the spikes in activity, we determined that five events were consistently correlated with higher levels of malware mentions:
Malware collections are multiple samples of different malware sold as a set, usually advertised as a collection of hacking tools that potential buyers can play around with to test which ones they like best. Of the references to malware collections found within the Recorded Future platform, the malware sold together were usually old or open-sourced malware already available on multiple forums. Because multiple malware variants were named in a collection’s sales post, the post showed up in multiple different forum search queries, and individuals searching for disparate families of malware converged on that post.
For example, njRAT’s greater number of references in June 2018 was correlated with a sales post for a “50 RAT Bundle” on now-defunct Dream Market, which contained builders for both families of malware. DarkComet and njRAT were both featured in a “561 RATs Collection” on a different forum in that same month. In addition, bundling tools that worked together also resulted in a spike in references. A post made on May 22, 2018 claimed that by using Xrumer (the forum spammer) alongside XEvil (a captcha checker), the actor was able to bypass over 8,400 captchas on forums, and most posts referencing Xrumer that were collected by Recorded Future in that month also referenced XEvil.
Malware vendors on Russian- and English-language sites tended to either post comments on the original sales thread or create new sales threads containing updates to the original malware variant. By updating the malware, vendors could update their sales posts and create higher post exposure within the forum to attract additional buyers. This sales post update would, on many occasions, cause the sales thread to get bumped to the forum’s front page. This is because many underground forums either sort forum threads by most recently active, or have a section advertising recently active comments or threads. GandCrab regularly announced updates to their namesake malware in this format.
In addition, the spike in activity correlated with updating malware not only applied to proprietary malware vendors, but also to open source developers and vendors of cracked goods. For vendors of cracked goods, having a newer version of their product differentiated them from other vendors selling the same good. For open source developers, fixing issues and adding new features to an open source tool that they could then share on malware forums allowed them to sell their version of the tool or develop their reputation on a certain forum. For example, while njRAT is openly available on GitHub, multiple forums contained posts of different “njRAT editions” either being sold or given away for free.
A cracked version of malware, like a cracked version of software, is malware that has been altered to allow for possible unauthorized copying and utilization of previously proprietary programs, breaching the terms of service with the original vendor. Eight malware variants named in the top 10 within this report had been cracked and sold on an underground forum this past year. The two families of malware that had not been sold in cracked versions are njRAT (which is open-sourced and does not need to be cracked) and GandCrab.
According to our research, we assess with medium confidence that there were four primary drivers that caused chatter around a family of cracked malware to grow. First, a cracked variant of malware allowed more individuals to access the malware, especially if the malware was initially only sold on a closed forum or a forum with limited access. Second, vendors who crack malware with the intention of selling it had an incentive to advertise their malware on multiple forums, as cracking another forum member’s malware can cause the new vendor to be banned from the initial forum. Third, altering malware to crack it, or altering the malware after it had been cracked, allowed vendors to tailor the additional functionality of the malware to suit audiences of different geographic regions. For example, multiple versions of Predator the Thief, an originally Russian malware family, had been cracked and altered in the past year to be tailored to an English audience.
Finally, the cracked version of the software could cause the original vendor to update their code to make the tool more appealing than the cracked, now more readily available, previous version. For example, a builder for AZORult, named “Gazorp,” was advertised on underground forums in September 2019 by an individual not endorsed by the original AZORult vendor. Gazorp created identical builds of AZORult 3.0, which was possible only because the source code for AZORult’s admin panel had been leaked earlier, alongside an older, legitimate AZORult builder. On October 4, 2018, the original author of AZORult, CrydBrox, updated their version of the stealer. During September and October 2018, the number of references for AZORult were far higher than in previous and later months. December 2018 saw a dip in references — during this time, CrydBrox announced that they were no longer selling or updating AZORult. AZORult’s reference count in January 2019 then spiked again, when the cracked version of AZORult 3.3 was released.
Ultimately, cracked malware is detrimental for both the original malware developer and for potential victims, as the original malware developer loses control of both the malware and any profits based on its sale. Additionally, because more individuals use the malware, this results in additional versions of the malware with different functionality, which makes defending against or hunting for the malware more difficult.
Of the 10 families of malware analyzed, the only families of proprietary malware still updated by the original author were GandCrab, Imminent Monitor, and WARZONE RAT (although the majority of WARZONE RAT and Imminent Monitor mentions on underground forums referenced cracked versions). While GandCrab has not yet been cracked, multiple vendors had attempted to sell their own ransomware under the GandCrab name. This grew the number of users who claimed to be using the malware, and provided incentives for the scammer selling the malware to put their sales post on multiple forums. For example, 360CN reported on a Chinese underground forum member selling a Chinese GandCrab ransomware builder. According to 360CN, the builder did create ransomware with the correct amount of ransom and Bitcoin address specified by a user, but also installed a cryptominer on the buyer’s machine.
Similarly to how cybersecurity blue teams and defenders of enterprise networks use news articles or security research as inspiration to strengthen defenses, cybercriminals use the same news articles or research as inspiration for how to breach defenses. In fact, almost all GandCrab posts on non-Russian forums were reposts of security news articles related to GandCrab updates, analysis reports on GandCrab, and decryptors for older versions of GandCrab. GandCrab also used news articles of their namesake malware to further advertise their product.
A malware family usually did not have a significant number of mentions simply through a single vendor posting about the malware. Our data demonstrated that all of the 10 malware variants analyzed in this piece had an engaged set of underground community members. njRat, AZORult, Predator the Thief, SpyNote, NLBrute, Xrumer, DarkComet, Imminent Monitor, and WARZONE RAT all have tutorials readily available on YouTube, created by security researchers and hackers alike. These tutorials make setup easier for newer buyers, allowing them to follow a demonstration of how to set up the administrator panel, or how to infect another machine for “research purposes.” In addition, buyers who used a specific malware family often became advocates for the malware by thanking the vendor publicly, posting a positive review of the vendor, or even posting about the malware on other forums or threads.
Because forum members could also post comments on the initial public sales thread, the vendor was also incentivized to provide customer service if a buyer posted a question.
Conversely, a majority of underground forums provided members with methods to report vendors of fake or faulty malware. This gave malware vendors an incentive to provide help to newer users or provide bug fixes and updates, while also allowing community members to ban less reputable vendors, whose posts would then either be deleted or die out.
The ultimate question that an analyst has to address when conducting forum-related research is how a malware family discussed on an underground forum actually makes an impact on a defender’s endpoint. In a few cases, malware advertised on underground forums will have little impact on a defender. For example, Xrumer is strictly a forum- or comment-based search engine optimization tool — while it can become a nuisance for any web application admin who has commenting enabled on their site, this tool is unlikely to do more damage than other tools specifically targeting web applications. In other cases, the malware is not effective without a delivery vehicle. For example, most RATs or ransomware require another vehicle of delivery, such as an exploit kit, phishing email, or brute-forced credentials, to install the malware on a victim host before it can be run. Additionally, many of the top 10 pieces of malware were advertised with crypters, as the vendors understood that their products could be easily detected with antivirus software.
However, some of the malware has been known to be used in longer campaigns using more sophisticated malware. For example, NLBrute was used to create the initial infection vector for actors to deliver SamSam ransomware, and APT-C-36 was seen utilizing Imminent Monitor in February 2019.
Insikt Group assesses with medium confidence that a majority of the malware analyzed in this report can be classified as a low to moderate threat to a customer environment, and that due to the number of underground forum members sharing, deploying, and providing reviews about the malware, the malware is likely seen in the wild with high frequency.
To measure the number of samples deployed in the wild, Insikt Group created VirusTotal queries for samples that flagged specific antivirus names for the malware families in the table below, from March 2018 onwards. The following queries likely represent a fraction of the overall samples in the wild — this is due to the number of defenders that will use VirusTotal over a separate malware analysis engine (if utilizing one at all), as well as tendencies for antivirus companies to tag malware as “general” or “malicious” instead of by the name of the malware. Insikt Group also used Shodan and measured the number of potential targets for NLBrute (which is used for initial infection and is not dropped on a victim host) and certain GandCrab campaigns. Each hash likely represents a malware sample dropped on at least one victim host, adding up to a minimum number of 417,163 infections in the last year.
|njRat||njRat samples submitted to VirusTotal after May 1, 2018||177,110 hashes|
|Predator the Thief||Predator the Thief samples submitted to VirusTotal after May 1, 2018||12,180 hashes|
|SpyNote||SpyNote android samples submitted to VirusTotal after May 1, 2018||113 hashes|
|AZORult||AZORult files submitted to VirusTotal after May 1, 2018||3,480 hashes|
|GandCrab||GandCrab samples submitted to VirusTotal after May 1, 2018||123,950 separate hashes|
|DarkComet||DarkComet samples submitted to VirusTotal after May 1, 2018||10,670 separate hashes|
|Imminent Monitor||Imminent Monitor samples submitted to VirusTotal after May 1, 2018||2,770 separate hashes|
|WARZONE RAT2||Ave Maria samples submitted to VirusTotal after May 1, 2018||2,500 separate hashes|
As seen in the table in the “Network Defense Recommendations” section below, Insikt Group also separated the malware by unique delivery mechanism. In the last year, every malware family had at least one associated campaign, used by either a criminal group or a state actor.
Using a data-driven approach, Insikt Group has ascertained the most popular malware and malware categories in the past year so that companies may better triage their threats and delegate their security resources. A majority of these malware variants were created over a year prior to this report, and some tools are also dual-use or open source tools, greatly underscoring the importance of cooperation between the SOC and internal penetration testing teams to ensure that enterprises are protected both from commodity malware and tools regularly used by security researchers. Tracking the growth of malware mentions over time also gave our team more insight into the habits of malware vendors and buyers, and what particular activity on underground forums makes certain malware more successful than others.
While Insikt Group realizes that most mature security teams will already be defending against most of these top 10, Recorded Future recommends that smaller companies with growing teams use this report to establish a patching baseline off the malware and vulnerabilities mentioned in the “Network Defense Recommendations” below and throughout this report. By first defending against the highest-frequency attacks, security teams can then focus on the more mature, quieter threats that employ defensive evasion or longer-term persistence.
Finally, Insikt Group ascertains with high confidence that the top tools on underground forums will continue to change. While this research has exclusively focused on trends in the last year, forum activity in the last seven months suggests that new tools are emerging. While the six families of malware in the top 10 remain popular, spikes in activity surrounding newly emerging Raccoon Stealer and new versions of Hawkeye Keylogger (during April 2019 and June 2019 respectively) have occurred, as well as gradual increases in comments about Gh0st RAT (an open source RAT first created by Chinese developers in 2008) and Star RAT (another Chinese RAT whose source code has been widely shared on Chinese underground forums).
Insikt Group recommends that defenders continue to monitor the underground to find their new baseline of high-frequency, low-medium impact tools to defend against. If defenders have ascertained that they are consistently targeted by actors in specific geographic regions, or actors who would likely be active on forums of a certain language, Insikt Group further recommends that defenders focus on finding their baseline by monitoring the tools most referenced on those specific forums.
Because most malware analyzed in this piece is not effective without a delivery vehicle, Insikt Group researchers separated the malware in the table below by delivery mechanism for campaigns using the malware reported on between 2018 and 2019 to ascertain vulnerabilities surrounding the malware. Insikt Group then found the vulnerabilities associated with the delivery mechanisms (see Appendix A).
|njRat||Malspam, Removable Drives|
|Predator the Thief||Malspam (CVE-2018-20250)|
|AZORult||RIG EK, Fallout EK|
|NLBrute||Delivery Mechanism for Other Payloads via Brute-Forcing RDP|
|GandCrab||MySQLdb Scanning, RIG EK, Phishing, Magnitude EK, Grandsoft EK|
|Xrumer||Delivery Mechanism for Spam|
|DarkComet||Malspam, Phishing (CVE-2014-6352)|
|Imminent Monitor||Phishing (CVE-2017-11882), Spearphishing|
|WARZONE RAT||Phishing (CVE-2017-11882)|
Editor’s Note: EK = Exploit Kit.
Insikt Group recommends that organizations conduct the following measures outlined in this section:
1Insikt Group defines “mentions” as a Recorded Future “reference.” This is either a post that contains an exact text match of an entity (in any language that Recorded Future analyzes), or a text match of a synonym, nickname, or abbreviation of that entity.
2Insikt Group assesses with high confidence that Ave Maria stealer is WARZONE RAT, based on previous reporting, C2 similarity, execution flow similarity, and Ave Maria RAT YARA rule matches on cracked WARZONE RAT samples found by Recorded Future. Additionally, samples from previous reporting on Ave Maria RAT found in the wild, as well as cracked versions of WARZONE RAT found on underground forums by Recorded Future, had a TLSH hashing score of 35, a score reserved for almost identical files.