IT threat evolution Q1 2020 – 10 minute mail

Targeted attacks and malware campaigns

Operation AppleJeus: the sequel

In 2018, we published a report on Operation AppleJeus, one of the more notable campaigns of the threat actor Lazarus, currently one of the most active and prolific APT groups. One notable feature of this campaign was that it marked the first time Lazarus had targeted macOS targets, with the group inventing a fake company in order to deliver its manipulated application and exploit the high level of trust among potential victims.

Our follow-up research revealed significant changes to the group’s attack methodology. To attack macOS victims, Lazarus has developed homemade macOS malware and added an authentication mechanism to deliver the next stage payload very carefully, as well as loading the next-stage payload without touching the disk. In addition, to attack Windows victims, the group has elaborated a multi-stage infection procedure and made significant changes to the final payload. We believe Lazarus has been more careful in its attacks since the release of Operation AppleJeus and has employed a number of methods to avoid detection.

We identified several victims as part of our ongoing research, in the UK, Poland, Russia and China. Moreover, we were able to confirm that several of the victims are linked to cryptocurrency business organizations.

Roaming Mantis turns to SMiShing and enhances anti-researcher techniques

Kaspersky continues to track the Roaming Mantis campaign. This threat actor was first reported in 2017, when it used SMS to distribute its malware to Android devices in just one country – South Korea. Since then, the scope of the group’s activities has widened considerably. Roaming Mantis now supports 27 languages, targets iOS as well as Android and includes cryptocurrency mining for PCs in its arsenal.

Roaming Mantis is strongly motivated by financial gain and is continuously looking for new targets. The group has also put a lot of effort into evading tracking by researchers, including implementing obfuscation techniques and using whitelisting to avoid infecting researchers who navigate to the malicious landing page. While the group is currently applying whitelisting only to Korean pages, we think it is only a matter of time before Roaming Mantis implements this for other languages.

Roaming Mantis has also added new malware families, including Fakecop and Wroba.j. The actor is still very active in using ‘SMiShing‘ for Android malware distribution. This is particularly alarming, because it means that the attackers could combine infected mobile devices into a botnet for malware delivery, SMiShing, and so on. In one of the more recent methods used by the group, a downloaded malicious APK file contains an icon that impersonates a major courier company brand: the spoofed brand icon is customized for the country it targets – for example, Sagawa Express for Japan, Yamato Transport and FedEx for Taiwan, CJ Logistics for South Korea and Econt Express for Russia.

WildPressure on industrial networks in the Middle East

In March, we reported a targeted campaign to distribute Milum, a Trojan designed to gain remote control of devices in target organizations, some of which operate in the industrial sector. We detected the first signs of this operation, which we have dubbed WildPressure, in August 2019; and the campaign remains active.

The Milum samples that we have seen so far do not share any code similarities with any known APT campaigns. All of them allow the attackers to control infected devices remotely: letting them download and execute commands, collect information from the compromised computer and send it to the C2 server and install upgrades to the malware.

Attacks on industrial targets can be particularly devastating. So far, we haven’t seen evidence that the threat actor behind WildPressure is trying to do anything beyond gathering data from infected networks. However, the campaign is still in development, so we don’t yet know what other functionality might be added.

To avoid becoming a victim of this and other targeted attacks, organizations should do the following.

  • Update all software regularly, especially when a new patch becomes available.
  • Deploy a security solution with a proven track record, such as Kaspersky Endpoint Security, that is equipped with behavior-based protection against known and unknown threats, including exploits.
  • On top of endpoint protection, implement a corporate-grade security solution designed to detect advanced threats against the network, such as Kaspersky Anti Targeted Attack Platform.
  • Ensure staff understand social engineering and other methods used by attackers and develop a security culture within in the organization.
  • Provide your security team with access to comprehensive cyberthreat intelligence, such as Kaspersky APT Intelligence Reporting.

TwoSail Junk

On January 10, we discovered a watering-hole attack that utilized a full remote iOS exploit chain to deploy a feature-rich implant named LightSpy. Judging by the content of the landing page, the site appears to have been designed to target users in Hong Kong.

Since then, we have released two private reports on LightSpy, available to customers of Kaspersky Intelligence Reporting (please contact [email protected] for further information).

We are temporarily calling the APT group behind this implant TwoSail Junk. Currently, we have hints from known backdoor callbacks to infrastructure about clustering this campaign with previous activity. We are also working with fellow researchers to tie LightSpy to prior activity from a well-established Chinese-speaking APT group, previously reported (here and here) as Spring Dragon (aka Lotus Blossom and Billburg(Thrip)), known for its Lotus Elise and Evora backdoors.

As this LightSpy activity was disclosed publicly by fellow researchers from Trend Micro, we wanted to contribute missing information to the story without duplicating content. In addition, in our quest to secure technologies for a better future, we have reported this malware and activity to Apple and other relevant companies.

Our report includes information about the Android implant, including its deployment, spread and support infrastructure.

A sprinkling of Holy Water in Asia

In December, we discovered watering-hole websites that were compromised to selectively trigger a drive-by download attack with fake Adobe Flash update warnings.

This campaign, which has been active since at least May 2019, targets an Asian religious and ethnic group. The threat actor’s unsophisticated but creative toolset, which has evolved greatly and may still be in development, makes use of Sojson obfuscation, NSIS installer, Python, open-source code, GitHub distribution, Go language and Google Drive-based C2 channels.

The threat actor’s operational target is unclear because we haven’t been able to observe many live operations. We have also been unable to identify any overlap with known APT groups.

Threat hunting with Bitscout

In February, Vitaly Kamluk, from the Global Research and Analysis Team at Kaspersky, reported on a new version of Bitscout, based on the upcoming release of Ubuntu 20.04 (scheduled for release in April 2020).

Bitscout is a remote digital forensics tool that we open-sourced about two and a half years ago, when Vitaly was located in the Digital Forensics Lab at INTERPOL. Bitscout has helped us in many cyber-investigations. Based on the widely popular Ubuntu Linux distribution, it incorporates forensics and malware analysis tools created by a large number of excellent developers around the world.

Here’s a summary of the approach we use in Bitscout

  • Bitscout is completely FREE, thereby reducing your forensics budget.
  • It is designed to work remotely, saving time and money that would otherwise be spent on travel. Of course, you can use the same techniques locally.
  • The true value lies not in the toolkit itself, but in the power of all the forensic tools that are included.
  • There’s a steep learning curve involved in mastering Bitscout, which ultimately reinforces the technical foundations of your experts.
  • Bitscout records remote forensics sessions internally, making it perfect for replaying and learning from more experienced practitioners or using as evidential proof of discovery.
  • It is fully open source, so you don’t need to wait for the vendor to implement a patch or feature for you: you are free to reverse-engineer and modify any part of it.

We have launched a project website, bitscout-forensics.info, as the go-to destination for those looking for tips and tricks on remote forensics using Bitscout.

Hunting APTs with YARA

In recent years, we have shared our knowledge and experience of using YARA as a threat hunting tool, mainly through our training course, ‘Hunting APTs with YARA like a GReAT ninja’, delivered during our Security Analyst Summit. However, the COVID-19 pandemic has forced us to postpone the forthcoming SAS.

Meanwhile, we have received many requests to make our YARA hands-on training available to more people. This is something we are working on and hope to be able to provide soon as an online training experience. Look out for updates on this by following us on Twitter – @craiu, @kaspersky.

With so many people working from home, and spending even more time online, it is also likely the number of threats and attacks will increase. Therefore, we decided to share some of the YARA experience we have accumulated in recent years, in the hope that all of you will find it useful for keeping threats at bay.

If you weren’t able to join the live presentation, on March 31, you can find the recording here.

We track the activities of hundreds of APT threat actors and regularly highlight the more interesting findings here. However, if you want to know more, please reach out to us at [email protected]

Other security news

Shlayer Trojan attacks macOS users

Although many people consider macOS to be safe, there are cybercriminals who seek to exploit those who use this operating system. One malicious program stands out – the Shlayer Trojan. In 2019, Kaspersky macOS products blocked this Trojan on every tenth device, making this the most widespread threat to people who use macOS.

Shlayer is a smart malware distribution system that spreads via a partner network, entertainment websites and even Wikipedia. This Trojan specializes in the installation of adware – programs that feed victims illicit ads, intercepting and gathering their browser queries and modifying search results to distribute even more advertising messages.

Shlayer accounted for almost one-third of all attacks on macOS devices registered by Kaspersky products between January and November last year – and nearly all other top 10 macOS threats were adware programs that Shlayer installs.

The infection starts with an unwitting victim downloading the malicious program. The criminals behind Shlayer set up a malware distribution system with a number of channels leading their victims to download the malware. Shlayer is offered as a way to monetize websites in a number of file partner programs, with relatively high payment for each malware installation made by users in the US, prompting over 1,000 ‘partner sites’ to distribute Shlayer. This scheme works as follows: a user looks for a TV series episode or a football match, and advertising landing pages redirect them to fake Flash Player update pages. From here, the victim downloads the malware; and for each installation, the partner who distributed links to the malware receives a pay-per-install payment.

Other schemes that we saw led to a fake Adobe Flash update page that redirected victims from various large online services with multi-million audiences, including YouTube, where links to the malicious website were included in video descriptions, and Wikipedia, where such links were hidden in article references. People that clicked on these links would also be redirected to the Shlayer download landing pages. Kaspersky researchers found 700 domains containing malicious content, with links to them on a variety of legitimate websites.

Almost all the websites that led to a fake Flash Player contained content in English. This corresponds to the countries where we have seen most infections – the US (31%), Germany (14%), France (10%) and the UK (10%).

Blast from the past

Although many people still use the term “virus” to mean any malicious program, it actually refers specifically to self-replicating code, i.e., malicious code that copies itself from file to file on the same computer. Viruses, which used to dominate the threat landscape, are now rare. However, there are some interesting exceptions to this trend and we came across one recently – the first real virus we’ve seen in the wild for some time.

The virus, called KBOT, infects the victim’s computer via the internet, a local network, or infected external media. After the infected file is launched, the malware gains a foothold in the system, writing itself to Startup and the Task Scheduler, and then deploys web injects to try to steal the victim’s bank and personal data. KBOT can also download additional stealer modules that harvest and send to the Command-and-Control (C2) server comprehensive information about the victim, including passwords/logins, crypto-wallet data, lists of files and installed applications, and so on. The malware stores all its files and stolen data in a virtual file system, encrypted using the RC6 algorithm, making it hard to detect.

Cybercriminals exploiting fears about data breaches

Phishers are always on the lookout for hot topics that they can use to hook their victims, including sport, politics, romance, shopping, banking, natural disasters and anything else that might entice someone into clicking on a link or malicious file attachment.

Recently, cybercriminals have exploited the theme of data leaks to try to defraud people. Data breaches, and the fines imposed for failing to safeguard data, are now a staple feature of the news. The scammers posed as an organization called the “Personal Data Protection Fund” and claim that the “US Trading Commission” had set up a fund to compensate people whose personal data had been exposed.

However, in order to get the compensation, the victims are asked to provide a social security number. The scammers offer to sell a temporary SSN to those who don’t have one.

Even if the potential victim enters a valid SSN, they are still directed to a page asking them to purchase a temporary SSN.

You can read the full story here.

… and coronavirus

The bigger the hook, the bigger the pool of potential victims. So it’s no surprise that cybercriminals are exploiting the COVID-19 pandemic. We have found malicious PDF, MP4 and DOCX files disguised as information about the coronavirus. The names of the files suggest they contain video instructions on how to protect yourself, updates on the threat and even virus detection procedures. In fact, these files are capable of destroying, blocking, modifying or copying data, as well as interfering with the operation of the computer.

The cybercriminals behind the Ginp banking Trojan recently developed a new campaign related to COVID-19. After receiving a special command, the Trojan opens a web page called Coronavirus Finder. This provides a simple interface that claims to show the number of people nearby who are infected with the virus and asks you to pay a small sum to see their location.

The Trojan then provides a payment form.

Then … nothing else happens – apart from the criminals taking your money. Data from the Kaspersky Security Network suggests that most users who have encountered Ginp are located in Spain. However, this is a new version of Ginp that is tagged “flash-2”, while previous versions were tagged “flash-es12”. So perhaps the lack of “es” in the tag of the newer version means the cybercriminals are planning to expand their campaign beyond Spain.

We have also seen a number of phishing scams where cybercriminals pose as bona fide organizations to trick people into clicking on links to fake sites where the scammers capture their personal information, or even ask them to donate money.

If you’ve ever wanted to know why it’s so easy for phishers to create spoof emails, and what efforts have been made to make it harder for them, you can find a good overview of the problems and potential solutions here.

Cybercriminals are also taking the opportunity to attack the information infrastructure of medical facilities, clearly hoping that the overload on IT services will provide them with an opportunity to break into hospital networks, or are attempting to extort money from clinical research companies. In an effort to ensure that IT security isn’t something that medical teams have to worry about, we’re offering medical institutions free six-month licenses for our core solutions.

In February, we reported an unusual malware campaign in which cybercriminals were spreading the AZORult Trojan as a fake installer for ProtonVPN.

The aim of the campaign is to steal personal information and crypto-currency from the victims.

The attackers created a spoof copy a VPN service’s website, which looks like the original but has a different domain name. The criminals spread links to the domain through advertisements using different banner networks – a practice known as malvertizing. When someone visits a phishing website, they are prompted to download a free VPN installer for Windows. Once launched, this drops a copy of the AZORult botnet implant. This collects the infected device’s environment information and reports it to the server. Finally, the attackers steal crypto-currency from locally available wallets (Electrum, Bitcoin, Etherium and others), FTP logins, and passwords from FileZilla, email credentials, information from locally installed browsers (including cookies), credentials from WinSCP, Pidgin messenger and others.

AZORult is one of the most commonly bought and sold stealers on Russian forums due to its wide range of capabilities. The Trojan is able to harvest a good deal of data, including browser history, login credentials, cookies, files and crypto-wallet files; and can also be used as a loader to download other malware.

Distributing malware under the guise of security certificates

Distributing malware under the guise of legitimate software updates is not new. Typically, cybercriminals invite potential victims to install a new version of a browser or Adobe Flash Player. However, we recently discovered a new approach: visitors to infected sites were informed that some kind of security certificate had expired.

They were offered an update that infected them with malware – specifically the Buerak downloader and Mokes backdoor.

We detected the infection on variously themed websites – from a zoo to a store selling auto parts. The earliest infections that we found date back to January 16.

Mobile malware sending offensive messages

We have seen many mobile malware apps re-invent themselves, adding new layers of functionality over time. The Faketoken Trojan offers a good example of this. Over the last six years, it has developed from an app designed to capture one-time passcodes, to a fully-fledged mobile banking Trojan, to ransomware. By 2017, Faketoken was able to mimic many different apps, including mobile banking apps, e-wallets, taxi service apps and apps used to pay fines and penalties – all in order to steal bank account data.

Recently, we observed 5,000 Android smartphones infected by Faketoken sending offensive text messages. SMS capability is a standard feature of many mobile malware apps, many of which spread by sending links to their victims’ contacts; and banking Trojans typically try to make themselves the default SMS application, in order to intercept one-time passcodes. However, we had not seen one become a mass texting tool.

The messages sent by Faketoken are charged to the owner of the device; and since many of the infected smartphones we saw were texting a foreign number, the cost was quite high. Before sending any messages, the Trojan checks to see if there are sufficient funds in the victim’s bank account. If there are, Faketoken tops up the mobile account sending any messages.

We don’t yet know whether this is a one-off campaign or the start of a trend. To avoid becoming a victim of Faketoken, download apps only from Google Play, disable the downloading of apps from other sources, don’t follow links from messages and protect your device with a reputable mobile security product.

The use and abuse of the Android AccessibilityService

In January, we reported that cybercriminals were using malware to boost the rating of specific apps, to increase the number of installations.

The Shopper.a Trojan also displays advertising messages on infected devices, creates shortcuts to advertising sites and more.

The Trojan opens Google Play (or other app store), installs several programs and writes fake user reviews about them. To prevent the victim noticing, the Trojan conceals the installation window behind an ‘invisible’ window. Shopper.a gives itself the necessary permissions using the Android AccessibilityService. This service is intended to help people with disabilities use a smartphone, but if a malicious app obtains permission to use it, the malware has almost limitless possibilities for interacting with the system interface and apps – including intercepting data displayed on the screen, clicking buttons and emulating user gestures.

Shopper.a was most widespread in Russia, Brazil and India.

You should be wary if an app requests access to the AccessibilityService but doesn’t need it. Even if the only danger posed by such apps comes from automatically written reviews, there is no guarantee that its creators will not change the payload later.

Everyone loves cookies – including cybercriminals

We recently discovered a new malicious Android Trojan, dubbed Cookiethief, designed to acquire root permissions on the victim’s device and transfer cookies used by the browser and the Facebook app to the cybercriminals’ C2 server. Using the stolen cookies, the criminals can gain access to the unique session IDs that websites and online services use to identify someone, thereby allowing the criminals to assume someone’s identity and gain access to online accounts without the need for a login and password.

On the C2 server, we found a page advertising services for distributing spam on social networks and messengers, which we think is the underlying motive in stealing cookies.

From the C2 server addresses and encryption keys used, we were able to link Cookiethief to widespread Trojans such as Sivu, Triada, and Ztorg. Usually, such malware is either planted in the device firmware before purchase, or it gets into system folders through vulnerabilities in the operating system and then downloads various applications onto the system.

Stalkerware: no place to hide

We recently discovered a new sample of stalkerware – commercial software typically used by those who want to monitor a partner, colleague or others – that contains functionality beyond anything we have seen before. You can find more information on stalkerware here and here.

MonitorMinor, goes beyond other stalkerware programs. Primitive stalkerware uses geo-fencing technology, enabling the operator to track the victim’s location, and in most cases intercept SMS and call data. MonitorMinor goes a few steps further: recognizing the importance of messengers as a means of data collection, this app aims to get access to data from all the popular modern communication tools.

Normally, the Android sandbox prevents direct communication between apps. However, if a superuser app has been installed, which grants root access to the system, it overrides the security mechanisms of the device. The developers of MonitorMinor use this to enable full access to data on a variety of popular social media and messaging applications, including Hangouts, Instagram, Skype and Snapchat. They also use root privileges to access screen unlock patterns, enabling the stalkerware operator to unlock the device when it is nearby or when they next have physical access to the device. Kaspersky has not previously seen this feature in any other mobile threat.

Even without root access, the stalkerware can operate effectively by abusing the AccessibilityService API, which is designed to make devices friendly for users with disabilities. Using this API, the stalkerware is able to intercept any events in the applications and broadcast live audio.

Our telemetry indicates that the countries with the largest share of installations of MonitorMinor are India, Mexico, Germany, Saudi Arabia and the UK.

We recommend the following tips to reduce the risk of falling victim to a stalker:

  • Block the installation of apps from unknown sources in your smartphone settings.
  • Never disclose the password or passcode to your mobile device, even with someone you trust.
  • If you are ending a relationship, change security settings on your mobile device, such as passwords and app location access settings.
  • Keep a check on the apps installed on your device, to see if any suspicious apps have been installed without your consent
  • Use a reliable security solution that notifies you about the presence of commercial spyware programs aimed at invading your privacy, such as Kaspersky Internet Security.
  • If you think you are being stalked, reach out to a professional organization for advice.
  • For further guidance, contact the Coalition against Stalkerware
  • There are resources that can assist victims of domestic violence, dating violence, stalking and sexual violence. If you need further help, please contact the Coalition against Stalkerware.


Temp Mails (https://tempemail.co/) is a new free temporary email addresses service. This service provide you random 10 minutes emails addresses. It is also known by names like: temporary mail, disposable mail, throwaway email, one time mail, anonymous email address… All emails received by Tempmail servers are displayed automatically in your online browser inbox.

APT trends report Q1 2020 – 10 minute mail

For more than two years, the Global Research and Analysis Team (GReAT) at Kaspersky has been publishing quarterly summaries of advanced persistent threat (APT) activity. The summaries are based on our threat intelligence research and provide a representative snapshot of what we have published and discussed in greater detail in our private APT reports. They are designed to highlight the significant events and findings that we feel people should be aware of.

This is our latest installment, focusing on activities that we observed during Q1 2020.

Readers who would like to learn more about our intelligence reports or request more information on a specific report are encouraged to contact ‘[email protected]’.

Given the exceptional situation the world is living in because of the COVID-19 pandemia, it is mandatory we to start with a summary of how APT groups have been abusing this topic for different types of attacks.

COVID-19 APT activity

Since the World Health Organization (WHO) declared the COVID-19 a pandemic, this topic has received increased attention from different attackers. Many of the phishing scams we’ve seen have been launched by cybercriminals trying to cash-in on people’s fears about the virus.  However, the list of attackers also includes APT threat actors such as Kimsuky, APT27, Lazarus or ViciousPanda who, according to OSINT, have used COVID-19-themed lures to target their victims. We recently discovered a suspicious infrastructure that could have been used to target health and humanitarian organizations, including the WHO. Even though the infrastructure cannot be attributed to any particular actor at the moment, and was registered before the COVID-19 crisis in June 2019, according to some private sources it might be related to the DarkHotel actor. However, we cannot confirm this information at the moment. Interestingly, some groups have used the current situation to try to soften their reputation by declaring that they would not target health organizations during the crisis.

There are different publications reporting activity related to other APT actors using this lure, but in general, we do not believe this implies a meaningful change in terms of TTPs other than using a trendy topic for luring victims. We are closely monitoring the situation.

The most remarkable findings

In January 2020, we discovered a watering-hole utilizing a full remote iOS exploit chain. This site appears to have been designed to target users in Hong Kong, based on the content of the landing page. While the exploits currently being used are known, the actor responsible is actively modifying the exploit kit to target more iOS versions and devices. We observed the latest modifications on February 7. The project is broader than we initially thought, supporting an Android implant, and probably supporting implants for Windows, Linux, and MacOS. For the time being, we are calling this APT group TwoSail Junk. We believe this is a Chinese-speaking group; it maintains infrastructure mostly within Hong Kong, along with a couple of hosts located in Singapore and Shanghai. TwoSail Junk directs visitors to its exploit site by posting links within the threads of forum discussions, or creating new topic threads ofтtheir own. To date, dozens of visits were recorded from within Hong Kong, with a couple from Macau. The technical details around the functionality of the iOS implant, called LightSpy, and related infrastructure, reveal a low-to-mid capable actor. However, the iOS implant is a modular and exhaustively functional iOS surveillance framework.

Russian-speaking activity

In January, a couple of recently compiled SPLM/XAgent modules were detected in an Eastern European telecoms company. The initial point of entry is unknown, as is their lateral movement within this organization. It has become rare to identify SPLM infections, compared to past levels of Sofacy activity, so it seems that portions of this network may have been infected for some time. In addition to these SPLM modules, Sofacy also deployed .NET XTUNNEL variants and their loaders. These 20KB XTUNNEL samples themselves seem minimal in comparison to past XTUNNEL samples, which weighed in at 1-2MB. This shift to C# by the long-standing Sofacy XTunnel codebase reminds us of Zebrocy’s practice of re-coding and innovating long-used modules in multiple languages.

Gamaredon, a well-known APT group that has been active since at least 2013, has traditionally focused on Ukrainian entities. In recent months we have observed a campaign, made up of different waves, that has also been reported by multiple researchers on different social networks. The attackers sent malicious documents with remote template injection, resulting in a multi-level infection scheme to deploy a malicious loader that periodically contacts a remote C2 to download additional samples. Based on past research, we know that the Gamaredon’s toolkit includes many different malware artefacts, developed to achieve different goals. These include scanning drives for specific system files, capturing screenshots, executing remote commands, downloading additional files and managing the remote machine with programs such as UltraVNC. In this case, we observed an interesting new second stage payload that includes spreading capabilities, that we call “Aversome infector”. This malware seems to have been developed to maintain a strong persistence in the target network and to move laterally by infecting Microsoft Word and Excel documents on external drives.

Chinese-speaking activity

CactusPete is a Chinese-speaking cyber-espionage group active since at least 2012 characterized by medium-level technical capabilities. Historically, this threat actor has targeted organizations within a limited range of countries – South Korea, Japan, the US and Taiwan. At the end of 2019 the group seemed to shift towards a heavier focus on Mongolian and Russian organizations. CactusPete offensive activity against the Russian defense industry and Mongolian government appears to be mostly delineated from its Russian-Mongolian commercial and border relationships. However, one bait exploit document dropping its Flapjack backdoor (tmplogon.exe, primarily focused on new Russian targets) is authored in Mongolian. The group’s broadening of techniques, exploit re-purposing, targeting shift and possible expansion suggests changes in the group’s resources and operations.

Rancor is a group that has been publicly reported since 2018, with connections to DragonOK. This actor traditionally had a focus on Southeast Asian targets, namely Cambodia, Vietnam and Singapore. We noted several updates to the group’s activity in the last few months, namely the discovery of a new variant of the Dudell malware that we are calling ExDudell, a new tool for bypassing UAC (User Account Control), and new infrastructure utilized in the attacks. Apart from this, we have also identified that the initial lure documents that were previously sent via mail, are now found in the Telegram Desktop directory, suggesting the group is possibly making a shift in its initial delivery method.

In 2019, we detected activity by an unknown actor at the time deploying watering holes on websites representing Tibetan interests, fooling victims into installing fake Adobe Flash updates hosted on a GitHub repository. Kaspersky thwarted the attack by coordinating a takedown of this repository with GitHub. After a brief period of inactivity, we detected a new round of watering holes featuring a renewed toolset. We decided to call the group behind this activity Holy Water.

The threat actor’s unsophisticated but creative toolset has been evolving a lot since the inception date, may still be in development, and leverages Sojson obfuscation, NSIS installer, Python, open-source code, GitHub distribution, Go language, as well as Google Drive-based C2 channels.

Middle East

We recently detected a new, ongoing data exfiltration campaign targeting victims in Turkey that started in February 2020. While StrongPity’s TTPs in terms of targeting, infrastructure and infection vector haven’t changed, we observed a somewhat peculiar change in the documents they attempt to exfiltrate. In this campaign, StrongPity updated its latest signature backdoor, named StrongPity2, and added more files to exfiltrate to its list of common Office and PDF documents, including Dagesh Pro Word Processor files used for Hebrew dotting, RiverCAD files used for river flow and bridge modelling, plain-text files, archives as well as GPG encrypted files and PGP keys.

In March, we discovered a targeted campaign to distribute Milum, a Trojan designed to gain remote control of devices in target organizations, some of which operate in the industrial sector. The first signs of this operation, which we have dubbed WildPressure, can be traced back to August 2019; still, the campaign remains active. The Milum samples we have seen so far do not share any code similarities with any known APT campaigns. The malware provides attackers with remote control over infected devices, allows downloading and executing commands, collecting and exfiltrating information and installing upgrades in the malware.

In late December 2019, Kaspersky Threat Attribution Engine detected a new variant of the Zerocleare wiper that had possibly been used in targeted attacks on energy sector targets in Saudi Arabia. This quarter, we identified a new variant of this wiper, called Dustman. It is similar to Zerocleare in terms of wiping and distribution, but changes in variables and technical names suggest this might have been in readiness for a new wave of attacks specifically targeting Saudi Arabia’s energy sector, based on messages embedded in the malware and the mutex created by it. The PDB file of the Dustman wiper suggested that this destructive code was the release edition and was ready for deployment in a target network. These changes coincided with the New Year holidays, during which many employees take time off to celebrate. Shamoon was delivered with similar timing in 2012 during Ramadan celebrations.

Southеast Asia and Korean Peninsula

A Lazarus campaign outlined by the Italian security company Telsy in November 2019 allowed us to find a connection to previous activity from the group targeting cryptocurrency businesses. The malware mentioned on Telsy’s blog is a first stage downloader that has been observed since mid-2018. We found that the second stage malware is a variant of Manuscrypt, uniquely attributed to Lazarus, deploying two types of payloads. The first is a manipulated Ultra VNC program, and the second is a multi-stage backdoor. This type of multi-stage infection procedure is typical of the Lazarus group’s malware, especially when using the Manuscrypt variant. In this campaign, our telemetry indicates that the Lazarus group attacked cryptocurrency businesses in Cyprus, the US, Taiwan and Hong Kong, and the campaign extended until the beginning of 2020.

Kimsuky, an actor we have been tracking since 2013, was especially active during 2019. In December, Microsoft took down 50 domains used by the group and filed a lawsuit against the attackers in a Virginia court. However, the group has continued its activity without significant changes. We recently discovered a new campaign where the actor used a decoy image themed around New Year’s greetings that delivers its old downloader with a new evolved next-stage payload designed to steal information that uses a new encryption method.

At the end of January, we stumbled upon a malicious script exploiting an Internet Explorer vulnerability, CVE-2019-1367. After closely examining the payload and finding connections with previous activity, we concluded that DarkHotel was behind this campaign, probably in progress since 2018. The campaign saw DarkHotel utilize a multi-stage binary infection phase using home-brewed malware. The initial infection creates a downloader which fetches another downloader to collect system information and fetch the final backdoor only for high-value victims. DarkHotel used a unique combination of TTPs in this campaign. The threat actor used diverse infrastructure to host malware and to control infected victims, including a compromised web server, a commercial hosting service, a free hosting service and a free source code tracking system. We were able to confirm targeted companies in South Korea and Japan in this campaign.

In March, researchers from Google revealed that a group of hackers used five zero-days to target North Koreans and North Korean-focused professionals in 2019. The group exploited flaws in Internet Explorer, Chrome, and Windows with phishing emails that carried malicious attachments or links to malicious sites, as well as watering-hole attacks. We were able to match two of the vulnerabilities – one in IE and one in Windows – to DarkHotel.

FunnyDream is a campaign that started in mid-2018, targeting high-profile entities in Malaysia, Taiwan and the Philippines, with the majority of victims in Vietnam. Our analysis revealed that it’s part of a wider campaign that stretches back a few years and targets governments, and specifically foreign organizations, of countries in Southeast Asia. The attacker’s backdoor downloads and uploads files from/to a C2, executes commands and runs new processes in the victim. It also collects information about other hosts on the network and is delivered to new hosts through remote execution utilities. The attacker also used an RTL backdoor and Chinoxy backdoor. The C2 infrastructure has been active since mid-2018 and domains show an overlap with the FFRAT malware family. In a number of cases, indications suggest the backdoor was delivered via a previous long-term compromise. The campaign is still active.

Operation AppleJeus was one of the more notable campaigns of Lazarus, and the first time the actor targeted macOS targets. Our January follow-up research revealed significant changes to the group’s attack methodology: homemade macOS malware and an authentication mechanism to carefully deliver the next-stage payload, as well as loading the next-stage payload without touching the disk. To attack Windows victims, the group has elaborated a multi-stage infection procedure and significantly changed the final payload. We believe that Lazarus has been more careful in its attacks since the release of Operation AppleJeus and has employed a number of methods to avoid detection. We identified several victims in the UK, Poland, Russia and China. Moreover, we were able to confirm that several of the victims are linked to cryptocurrency organizations.

Roaming Mantis is a financially motivated actor first reported in 2017, when it used SMS to distribute its malware to Android devices based in South Korea. Since then, the scope of the group’s activities has widened considerably, supporting 27 languages, targeting iOS as well as Android, and even mining cryptocurrency. The actor also added new malware families, including Fakecop and Wroba.j to its arsenal, and is still active using ‘SMiShing‘ for Android malware distribution. In a recent campaign it distributed malicious APKs masquerading as popular couriers and customized for the targeted countries, including Japan, Taiwan, South Korea and Russia.

Other interesting discoveries

TransparentTribe started using a new module named USBWorm at the beginning of 2019, as well as improving its custom .NET tool named CrimsonRAT. Based on our telemetry, USBWorm was used to infect thousands of victims, most of them located in Afghanistan and India, providing the attacker with the ability to download and execute arbitrary files, spread to removable devices and steal files of interest from infected hosts even those disconnected from the internet. As we previously reported, this group mainly focuses on military targets, which are usually compromised with Office documents armed with malicious VBA and open-source malware like Peppy RAT and CrimsonRAT. In its new campaign, which is still active, we noticed the group’s focus shift more towards targeting entities located in Afghanistan in addition to India. Transparent Tribe has also developed a new implant designed to infect Android devices, a modified version of the AhMyth Android RAT which is open source malware available on GitHub.

During the last months of 2019, we observed an ongoing campaign conducted by Fishing Elephant. The group continues to use both Heroku and Dropbox in order to deliver its tool of choice, AresRAT. We discovered that the actor incorporated a new technique into its operations that is meant to hinder manual and automatic analysis – geo-fencing and hiding executables within certificate files. During our research, we also detected a change in victimology that may reflect the current interests of the threat actor: the group is targeting government and diplomatic entities in Turkey, Pakistan, Bangladesh, Ukraine and China.

Final thoughts

While the threat landscape isn’t always full of “groundbreaking” events, when we cast our eyes back over the activities of APT threat actors, there are always interesting developments.  Our regular quarterly reviews are intended to highlight the key developments.

These are some of the main trends that we’ve seen this year so far.

  • It’s clear from the activities of various APT groups, including CactusPete, LightSpy, Rancor, Holy Water, TwoSail Junk and others that geo-politics continues to be an important driver of APT activity. This was also underlined this quarter by the UK National Cyber Security Centre laying responsibility for disruptive attacks on Georgia at the feet of Russia’s military intelligence service, indictments in the US of two Chinese nationals for laundering $100 million in cryptocurrency on behalf of North Korea and the alleged ‘catfishing’ of IDF soldiers by Hamas.
  • Financial gain remains a motive for some threat actors, as evidenced by the activities of Lazarus and Roaming Mantis.
  • Southeast Asia is the most active region in terms of APT activities, including established actors such as Lazarus, DarkHotel and Kimsuky, and newer groups such as Cloud Snooper and Fishing Elephant.
  • APT threat actors such as CactusPete, TwoSail Junk, FunnyDream, DarkHotel continue to exploit software vulnerabilities.
  • APT threat actors continue to include mobile implants in their arsenal.
  • APT threat actors such as (but not limited to) Kimsuky, Hades and DarkHotel, as well as opportunistic criminals, are exploiting the COVID-19 pandemic.

All in all, we see the continuous growth of activity in Asia and how some of the actors we called newcomers are now well established. On the other hand, the more traditional advanced actors seem to be more and more selective in their operations, probably following a change of paradigm. The use of mobile platforms for infections and the distribution of malware is on the rise. Every actor seems to have some artefacts for these platforms and in some campaigns they are the main target.

COVID-19 is clearly top of everyone’s minds at the moment and APT threat actors have also been seeking to exploit this topic in spear-phishing campaigns.  We do not believe this represents a meaningful change in terms of TTPs: they’re simply using it as a newsworthy topic to lure their victims. However, we are closely monitoring the situation.

As always, we would note that our reports are the product of our visibility into the threat landscape. However, it should be borne in mind that, while we strive to continually improve, there is always the possibility that other sophisticated attacks may fly under our radar.


Temp Mails (https://tempemail.co/) is a new free temporary email addresses service. This service provide you random 10 minutes emails addresses. It is also known by names like: temporary mail, disposable mail, throwaway email, one time mail, anonymous email address… All emails received by Tempmail servers are displayed automatically in your online browser inbox.

A look at the ATM/PoS malware landscape from 2017-2019 – 10 minute mail

From remote administration and jackpotting, to malware sold on the Darknet, attacks against ATMs have a long and storied history.  And, much like other areas of cybercrime, attackers only refine and grow their skillset for infecting ATM systems from year-to-year. So what does the ATM landscape look like as of 2020? Let’s take a look.

The world of ATM/PoS malware

ATM attacks aren’t new, and that’s not surprising. After all, what is one of the primary motives driving cyber criminals? Money. And ATMs are cash hubs—one successful attack can net you hundreds of thousands of dollars. In the past, even high-profile threat actors have made ATMs their prime target.

However, attacking ATMs is a bit different from traditional financial-related threats, like phishing emails or spoofed websites. That’s because ATMs operate in a unique space in the tech world: they’re still connected to the corporate networks but at the same time must be accessible to anyone that passes by. The resulting technical differences means the attack methods differ from those used for traditional endpoints.

ATMs also share several common characteristics that make them particularly vulnerable to attacks:

  • Traditional software that is part of the warranty offered by the vendors → If major changes occur that are not approved by the ATM vendor, including installing AV software, then sometimes this warranty is lost.
  • Regular use of outdated operating systems and the apps its runs on
  • Locations chosen in a way that provide access to as many customers as possible, including those in remote regions → These isolated locations often lack any reasonable physical security

Old software means unpatched vulnerabilities—ones criminals can exploit—and isolated areas makes it easier for criminals to gain physical access to the internal ports of the motherboard. This is especially typical for the old ATM machines located in many regions with low resources and no budgets for ATM upgrades.  When combined, ATMs become not only a highly profitable target—but an easy one.

From 2017 to 2019, there has been a marked increase in ATM attacks, due to a few families being particularly active. These target systems around the globe, regardless of the vendor, and have one of two goals: either stealing customers’ information or funneling funds directly from the bank.

Considering all of the above, we decided to delve further into what has been happening in the world of ATM/PoS malware for the last few years.

ATM/oOS malware attacks: by the numbers

To gain a closer look at ATM malware worldwide, we utilized the statistics processed by Kaspersky Security Network (KSN) over the course of the past three years globally.

Number of unique devices that encountered ATM/PoS malware, 2017-2019 (download)

The results showed that the number of unique devices protected by Kaspersky that encountered ATM/PoS (point-of-sale) malware at least once experienced a two-digit growth in 2018—and this number held steady, even increasing slightly, in 2019.

Geography of unique devices that encountered ATM/PoS malware, 2017 (download)

TOP 10 countries by number of unique devices that encountered ATM/PoS malware in 2017

Country Devices
1 Russian Federation 1016
2 Brazil 423
3 Vietnam 281
4 United States 148
5 India 137
6 Turkey 96
7 China 94
8 Germany 58
9 Philippines 53
10 Mexico 51

The ten countries that had the greatest number of unique devices affected by ATM/POS malware were relatively dispersed around the globe, with the highest number in Russia. Russia has had a long history of threat actors targeting financial institutions. For example, it was in 2017 that Kaspersky researchers  uncovered an ATM malware dubbed “ATMitch” that was gaining remote access control over ATMS at Russian banks. In addition, the relatively high rates in both Brazil and Mexico can be partially attributed to Latin and South America’s longstanding history as a hotspot of ATM malware.

Geography of unique devices that encountered ATM/PoS malware, 2018 (download)

TOP 10 countries by number of unique devices that encountered ATM/PoS malware in 2018

Country Devices
1 Russian Federation 1370
2 Brazil 753
3 Italy 537
4 United States 519
5 Vietnam 433
6 India 408
7 Thailand 369
8 Germany 277
9 Turkey 224
10 Iran 198

In 2018, the countries with the greatest number of ATM/PoS malware incidents recorded by unique devices remained distributed worldwide, but the countries remained similar to 2017, with the highest activity recorded in Russia and Brazil.

The overall increase in the number of devices affected can be attributed to both the reappearance of new ATM malware and the development of new families:

  • ATMJackpot first appeared in Taiwan back in 2016. It infects the banks’ internal networks, allowing it to withdraw funds directly from the ATM. ATMJackpot was able to reach thousands of ATMs.
  • WinPot was discovered at the beginning of 2018 in Eastern Europe and was designed to make the infected ATM automatically dispense all cash from its most valuable cassettes. Because of its time counter, its execution is time-dependent: if the targeted system’s time does not fall within the preset period during which the malware was programmed to work (e.g. March), WinPot silently stops operating without showing its interface.
  • Ice5 originated in Latin America. Its engineering tool is written in a scripting language that allows the attackers to achieve a significant level of manipulation over the infected ATMs. The initial infection occurs via the USB port.
  • ATMTest is a multi-stage infection in 2018. It requires console access to the ATM, meaning the attackers have to gain remote access to the bank’s networks. This malware was originally coded to steal money in rubles.
  • Peralta was an evolution of the infamous ATM malware project called Ploutus, which led to losses of $64,864,864.00 across 73,258 compromised ATMs. Both Peralta and Ploutus originated in Latin America.
  • ATMWizX was discovered in the fall of 2018 and dispenses all cash automatically, starting with the most valuable cassettes.
  • ATMDtruck also appeared in the fall of 2018 with indications that the first victims were in India. It collects enough information from the credit cards inputted into the infected ATM that it can actually clone them. It drops the malware “Dtrack”, which is a sophisticated spy tool.

Geography of unique devices that encountered ATM/PoS malware, 2019 (download)

TOP 10 countries by number of unique devices that encountered ATM/PoS malware in 2019

Country Devices
1 Russian Federation 2306
2 Iran 1178
3 Brazil 819
4 Vietnam 416
5 India 353
6 Germany 228
7 United States 220
8 Italy 197
9 Turkey 149
10 Mexico 114

This past year, the ten countries with the highest level of ATM/PoS malware activity remained the same, with only one change: Mexico once again entered the top ten, while Thailand left.

Overall, the total number of devices affected increased once again. In fact, ATM/PoS malware activity reached new levels by the spring of 2019 with a string of operations: ATMqot, ATMqotX, and ATMJaDi. ATMgot operates directly on the ATM using the dispenser to withdraw the maximum number of banknotes allowed; if it cannot do this, it will default to 20 notes. This malware also possesses anti-forensic techniques that allow it to delete traces of the infection from the ATMs, as well as some video files, which could potentially be used as part of video monitoring.

ATMJadi orginated in Latin America and is capable of cashing out ATMs. Since it’s a Java-based project, it’s platform-dependent—and thus highly targeted. In order to be installed, the attackers must gain access to the bank’s network. This suggests the attackers first compromise the bank’s infrastructure. But what’s perhaps most interesting is the false flag section with strings in the Russian language.

The problem of cyberattacks is compounded by the use of outdated and unpatched systems. That means that, even as new 2019 malware families were developed, the old ATM families from the previous years can still be used to launch successful attacks.

A look towards the future

ATM/PoS malware will only continue to evolve, and so, we will continue to monitor the ecosystem closely. We’ve already seen WinPot, first discovered in 2018, active this year in different parts of the world.

Latin America has long been known as a region of innovative cybercriminals who adopt techniques other region uses. It’s not surprising then that a new trend was recently discovered in development: an ATM MaaS project whereby a group in Latin America is attempting to sell ATM malware developed for each major vendor on the market. Projects like these provide further evidence that the world of ATM malware is still evolving, with cybercriminals continuously developing better attack strategies.

Our research has also shown that, beyond Latin America, countries in Europe and the APAC region are of particular interest to ATM attackers, as is the United States. This signifies that ATM malware is a truly global threat. After all, ATMs are located in nearly every country and few systems offer access to such massive amounts of fund.

How, then, can you protect your money? No matter how digital banking has become, ATMs are still an inevitable part of managing your funds. While you can’t control whether or not an ATM machine is attacked, by conscientiously monitoring your accounts and financial transactions, you can make sure suspicious activity is quickly identified and the proper channels duly notified. This should help mitigate the damage caused by any attack.

For financial institutions, staying secure requires a comprehensive, multi-step approach:

  1. Evaluate which attack vectors are more likely to be used and generate a threat model. This will depend, for example, on what network architecture is in place and where the ATM is installed – a place not controlled by your organization, such as a wall on the street, or an office under video surveillance, etc.
  2. Determine which ATMs are outdated or have an OS version that’s reaching the end of its vendor support. If you cannot replace the legacy devices, pay attention to this fact in your threat model and set the appropriate security solution settings, which do not affect the device’s productivity.
  3. Regularly conduct security assessments or pentests of ATMs to find possible cyberattack vectors. Kaspersky’s threat hunting service can also help you find sophisticated cybercriminals.
  4. Regularly review the physical safety of ATMs to detect abnormal elements implemented by attackers.
  1. If ATM configurations permit it, install a security solution that protects the devices from different attack vectors, such as Kaspersky Embedded Systems Security. If the device has extremely low system specs, the Kaspersky solution would still protect it with a Default Deny whitelisting scenario

PoS terminals are in many aspects similar to ATMs, but still possess a number of differences to be mindful of—and tackled accordingly. Apart from the steps mentioned above (which remain applicable), the following must be taken into account:

  1. Often more powerful when compared to an average ATM, Windows-based PoS terminals offer greater spaces for attackers’ maneuvering and are capable of running a broad range of modern malware and hacking tools. This makes implementation of multi-layered protection a must.
  2. While also residing in public spaces, they generally lack ATMs’ heavy armor. Therefore, they are more susceptible to direct attacks using unauthorized devices. This makes properly configured Device Control even more valuable.
  3. As they are frequently involved not only in financial, but also personal, data processing, this adds to their attractiveness for cyberattacks and also subjects them to more legislation. In combination with direct attack scenarios, implementation of file integrity monitoring and log inspection are mandatory, preferably in a way that allows tracking changes offline.
  4. Embedded systems should be protected not only by host-based security, but also by application of network-level security, such as Secure Web Gateways or Next-gen Firewalls capable of detecting and blocking unsolicited communications and other systems both inside and outside of the company’s infrastructure.


Temp Mails (https://tempemail.co/) is a new free temporary email addresses service. This service provide you random 10 minutes emails addresses. It is also known by names like: temporary mail, disposable mail, throwaway email, one time mail, anonymous email address… All emails received by Tempmail servers are displayed automatically in your online browser inbox.

Loncom packer: from backdoors to Cobalt Strike – 10 minute mail

The previous story described an unusual way of distributing malware under disguise of an update for an expired security certificate. After the story went out, we conducted a detailed analysis of the samples we had obtained, with some interesting findings. All of the malware we examined from the campaign was packed with the same packer, which we named Trojan-Dropper.NSIS.Loncom. The malware uses legitimate NSIS software for packing and loading shellcode, and Microsoft Crypto API for decrypting the final payload. Just as the earlier find, this one was not without its surprises, as one of the packaged samples contained software used by APT groups.

Primary analysis

Loncom utilizes NSIS for running shellcode contained in a file with a name that consists of numbers. In our example, the file is named 485101134:

Overview of the NSIS archive contents

Once the shellcode is unpacked to the hard disk and loaded into the memory, an NSIS script calculates the starting position and proceeds to the next stage.

What the shellcode does

Before proceeding to decrypt the payload, the shellcode starts decrypting itself piece by piece, using the following algorithm:

  • Find position for next 0xDEADBEEF dword.
  • Read dword: size of data to decrypt.
  • Read dword: first part of key.
  • Read dword: second part of key.
  • Find suitable key: check the numbers consequently, starting at 0, while xor(i, second part of key) != first part of key. This part is needed to hold up execution and prevent AV detection. After simplification, key = i = xor(first part, second part).
  • Decrypt next part of shellcode (xor), move on to it.

Decrypting the next part of the shellcode

Here’s the code that performs the algorithm described above:

After several such iterations of block decryption, the shellcode switches to active steps, loading libraries and retrieving the addresses of required functions with the help of the APIHashing technique. This helps avoid stating the names of requested functions directly, providing their hashes instead. When searching for functions by hash, a hash will be calculated for each element from the library export table until it matches the target.

Then, Loncom decrypts the payload contained in the same file as the shellcode and proceeds to run it. The payload is encrypted with an AES-256 block cipher. The decryption key is stated in the code, and the payload offset and size are passed from the NSIS script.

The main part of the shellcode: decrypting the payload

Unpacking

For automated Loncom unpacking, we need to find out how data is stored in the packed NSIS installers, obtain the payload offset and size from the NSIS script, and pull the key from the shellcode.

Unpacking the NSIS

After a brief analysis, we managed to find that the NSIS installers have the following structure:

  • an MZPE NSIS interpreter containing in its overlay the data to be processed: the flag, the signatures, the size of the unpacked header, and the total size of the data, and then the containers, i.e. the compressed data itself.
  • Containers in the following format: dword (data size):zlib_deflate(data). The 0th container has the header, the first container has our file with the shellcode and the payload, and the second one has the DLL with the NSIS plugin.
  • The header contains a table of operation codes for the NSIS interpreter, a string table and a language table.

As we have obtained the encrypted file, now all we need is to find the payload offset and size, and proceed to decrypting the payload and the shellcode.

NSIS data structure

As all arguments in the NSIS operation codes when using plugins are passed as strings, we need to retrieve from the header string table all strings that look like numbers within the logical limits: from 0 to (file size – shellcode size).
NSIS unpacking code:

To simplify determining the payload offset and size, we can recall the structure of the file with the shellcode: encrypted blocks are decrypted from the smallest address to the largest, top to bottom, and the payload is located above the shellcode. Thus, we can determine the position of the 0xDEADBEEF byte and consider it the end of the encrypted data (aligning as required, because AES is a block cipher).

Decrypting the shellcode

To decrypt the payload, we need to:

  • decrypt the shellcode blocks;
  • determine where the AES key is;
  • retrieve the key;
  • try to decrypt the payload for offsets received from the NSIS;
  • stop after obtaining the first two bytes = ‘MZ’.

Step one can be performed by slightly modifying the code that performs the decryption algorithm in IDA Pro. The key can be determined with the help of a simple regular expression: ‘xC7x45.(….)xC7x45.(….)xC7x45.(….)xC7x45.(….)xE8’ — “mov dword ptr” 4 times, then “call” (pseudocode in the main part of the shellcode).
The other steps do not require a detailed explanation. We will now describe the actual malware that was packed with Loncom.

What’s inside

Besides Mokes and Buerak, which we mentioned in the previous article, we noticed packed specimens of Backdoor.Win32.DarkVNC and Trojan-Ransom.Win32.Sodin families, also known as REvil and Sodinokibi. The first is a type of backdoor used for controlling an infected machine via the VNC protocol. The second is a ransomware that encrypts the victim’s information and threatens to publish it.
However, the most exciting find was the Cobalt Strike utility, used both by legal pentesters and by various APT groups. The command center of the sample that contained Cobalt Strike had previously been seen distributing CactusTorch, a utility for running shellcode present in Cobalt Strike modules, and the same Cobalt Strike packed with a different packer.

We continue monitoring Trojan-Dropper.NSIS.Loncom and hope to share new findings soon.

IOC

BB00BA9726F922E07CF243D3CCFC2B6E (Backdoor.Win32.DarkVNC)
EBE191BF77044961684DF51B88CA8D05 (Backdoor.Win32.DarkVNC)
4B4C98AC8F04680F7C529956CFE8519B (Trojan-Ransom.Win32.Sodin)
AEF8FBB5C64734093E78EB13E6FA7849 (Cobalt Strike)


Temp Mails (https://tempemail.co/) is a new free temporary email addresses service. This service provide you random 10 minutes emails addresses. It is also known by names like: temporary mail, disposable mail, throwaway email, one time mail, anonymous email address… All emails received by Tempmail servers are displayed automatically in your online browser inbox.

Holy water: ongoing targeted water-holing attack in Asia – 10 minute mail

On December 4, 2019, we discovered watering hole websites that were compromised to selectively trigger a drive-by download attack with fake Adobe Flash update warnings. This campaign has been active since at least May 2019, and targets an Asian religious and ethnic group.

The threat actor’s unsophisticated but creative toolset has been evolving a lot since the inception date, may still be in development, and leverages Sojson obfuscation, NSIS installer, Python, open-source code, GitHub distribution, Go language, as well as Google Drive-based C2 channels.

The threat actor’s operational target is not clear because, unfortunately, we haven’t been able to observe many live operations, and we couldn’t identify any overlap with known intrusion sets.

Thou shalt update plugins: attack synopsis

The watering holes have been set-up on websites that belong to personalities, public bodies, charities and organizations of the targeted group. At the time of writing, some of these websites (all hosted on the same server) are still compromised, and continue to direct selected visitors to malicious payloads:

Domain Description
*****corps.org Voluntary service program
*****ct.org Religious personality’s charity
*****policy.net Policy institute
*****che.com Religious personality
*****parliament.org Public body
*****ialwork.org Charity
*****nature.net Environmental conservation network
*****airtrade.com Fair trade organization

Upon visiting one of the watering hole websites, a previously compromised but legitimately embedded resource will load a malicious JavaScript. It’s hosted by one of the water-holed websites, and gathers information on the visitor. An external server (see Fig. 1) then ascertains whether the visitor is a target.

Fig. 1. Target validation service request.

If the visitor is validated as a target, the first JavaScript stage will load a second one, which in turn will trigger the drive-by download attack, showing a fake update pop-up (see Fig. 2).

Fig. 2. Warning generated by the second payload.

The visitor is then expected to fall into the update trap, and download a malicious installer package that will set up a backdoor.

For nothing is hidden that will not come to light: technical analysis

1st JavaScript stage

The first JavaScript stage is named (script|jquery)-css.js, and is obfuscated with the Chinese-language web service Sojson, version 4 (see Fig. 3).

Fig. 3. Sojson v4 JavaScript obfuscated one-liner.

The payload leverages the RTCPeerConnection API and ipify service to fingerprint visitors. The gathered data is sent to loginwebmailnic.dynssl[.]com through HTTP GET requests, in order to validate the visitor as a target:

https://loginwebmailnic.dynssl[.]com/all/content.php?jsoncallback=&lanip=&wanip=&urlpath=&_=

The JSON-formatted response, whose only key is “result”, can either be “t” or “f” (true or false). If the value is “f”, then nothing happens, while “t” will trigger the second JavaScript stage (see Fig. 4).

Fig. 4. First stage deobfuscated validation logic.

In a previous version of this first JavaScript script, an additional JavaScript payload was unconditionally loaded during the first stage, and proceeded with another branch of visitor validation and the second stage.

This other branch loaded scripts from root20system20macosxdriver.serveusers[.]com, and leveraged https://loginwebmailnic.dynssl[.]com/part/mac/contentmc.php URL to validate targets. The host and validation page names suggest this other branch may have been specifically targeting MacOS users, but we were unable to confirm this hypothesis.

2nd JavaScript stage

The second JavaScript stage is named (script|jquery)-file.js, and is obfuscated with Sojson version 5 (see Fig. 5).

Fig. 5. Nerve-breaking one-line obfuscation.

The payload leverages jquery.fileDownload to show a modal pop-up to the target. It offers visitors an update to Flash Player. No technical vulnerabilities are exploited: the threat actor relies on the target’s willingness to keep their system up to date. The deobfuscated JavaScript payload (see Fig. 6) reveals that the malicious update is hosted on GitHub.

Fig. 6. Malicious update source in second JavaScript payload.

GitHub FlashUpdate repository

The pop-up links to a PE executable hosted on github[.]com/AdobeFlash32/FlashUpdate. GitHub disabled this repository on February 14 after we reported it to them. However, the repository has been online for more than nine months, and thanks to GitHub’s commit history (see Fig. 7), we gained a unique insight into the attacker’s activity and tools.

Fig. 7. GitHub’s AdobeFlash32 commit history.

Four executables were hosted in AdobeFlash32/FlashUpdate on the last day it was still available:

  • An installer package, embedding a decoy legitimate Flash update and a stager.
  • Godlike12, a Go backdoor that implements a Google Drive based C2 channel.
  • Two versions of the open-source Stitch Python backdoor that the threat actor modified to add functionalities (persistence, auto-update, decoy download and execution).

Digging into the repository for older commits, we also discovered a previous fake update toolset: a C installer bundling the legitimate Flash installer and a vanilla Stitch backdoor, as well as a C++ infostealer that collects information about host computers (OS version, IP address, hostname) and sends them over HTTP/S.

Installer package

MD5 9A819F2CE060058745FF5374221ADA7C
Compilation date 2017-Jul-24 06:35:22
File type PE32 executable (GUI) Intel 80386, for MS Windows, Nullsoft Installer self-extracting archive
File size 4420 KB
File names flashplayer32ppi_xa_install.exe

This malicious update package is a NSIS installer version 3 that will drop and execute two other binaries:

  • FlashUpdate.exe, D59B35489CB88619415D175953CA5400, a legitimate Windows Flash Player installer from January 15 that is used as a decoy to trick the user into believing they actually set up a Flash update. As modern Adobe Flash installers ‘phone home’ to check for their own validity, this one will fail nowadays with a message stating that the installer is outdated or renamed, and will direct the user to the Adobe website.
  • Intelsyc.exe, the malicious payload (described below).

The installer is detected by Kaspersky endpoint protection heuristics as HEUR:Trojan.Win32.Tasker.gen.

Intelsyc Go stager

MD5 6DC5F8282DF76F4045F75FEA3277DF41
Compilation date 1970-Jan-01 00:00:00
File type PE32 executable (GUI) Intel 80386 (stripped to external PDB), for MS Windows
File size 5976 KB
File names flashplayer32ppi_xa_install.exe
C2 server adobeflash31_install.ddns[.]info
User Agent Go-http-client/1.1

The Go programmed Intelsyc implant is aimed at staging itself, downloading the Godlike12 backdoor (described below), and setting up persistence.

It will first retrieve /flash/sys.txt with HTTP GET on adobeflash31_install.ddns[.]info. The file contents may be used as a killswitch to stop any further deployment. If the content is “1” though, the implant will:

  • copy itself to C:/ProgramData/Intel/Intelsyc.exe;
  • establish persistence through schtasks [T1053] with a logon task named Intelsyc, run as system, and pointing to a previously created self copy;
  • download Godlike12 from github[.]com/AdobeFlash32/FlashUpdate, as C:ProgramDataAdobeflashdriver.exe;
  • establish Godlike12 persistence through a registry run key [T1060] named flashdriver in HKLMSOFTWAREMicrosoftWindowsCurrentVersionRun, and pointing to a previously downloaded backdoor.

The stager is detected by Kaspersky endpoint protection heuristics as UDS:DangerousObject.Multi.Generic, and may be misidentified as the GoRansom Go ransomware proof of concept by other endpoint protection products.

Source files paths in the code suggest this backdoor may have been developed on a Windows system.

Godlike12 Go backdoor

MD5 BEC4482890A89F0184B463C727709D53
Compilation date 1970-Jan-01 00:00:00
File type PE32 executable (GUI) Intel 80386 (stripped to external PDB), for MS Windows
File size 4436 KB
File names flashdriver.exe
C2 server Google Drive

This implant is written in Go language, and its C2 channel relies on file exchanges with a Google Drive space, through Google Drive’s HTTPS API v3. The implant probably leverages the gdrive Go source from GitHub, as it shares several identical code source paths with it.

Godlike12 is the name the threat actor gave to the Google Drive space connections from this implant. Source file paths in the code suggest this backdoor may have been developed on a GNU/Linux system. The not-so-common (less than 100 results in a popular search engine) /root/gowork GOPATH that some of this backdoor’s modules have been compiled from seems popular in Chinese-speaking communities, and may originate from a Chinese-authored tutorial on Go language.

Godlike12 first proceeds with host fingerprinting upon startup (hostname, IP address, MAC address, Windows version, current time). The result is encrypted, base64-encoded, stored in a text file at %TEMP%/[ID]-lk.txt, and uploaded to the remote Google Drive. The implant then regularly checks for a remote [ID]-cs.txt, that contains encrypted commands to execute, and stores encrypted command results in %TEMP%/[ID]-rf.txt to later upload them to the same Google Drive space. ID is the MD5 hash of the base64-encoded MAC address of the first connected network adapter, while TripleDES in ECB mode is used as an encryption algorithm. It is worth mentioning that once again, the encryption function seems to have been inspired from existing open-source code, which mainly appears popular in Chinese-language forums.

Godlike12 does not implement a persistence mechanism, as it is provided by the previous installer package. It is detected by Kaspersky endpoint protection heuristics as HEUR:Trojan.Win32.Generic.

With this implant being a month old at the time of writing (while being in use since at least October 2019), and other malicious update implants having been used before, it is possible that Godlike12-based operations were still a work in progress when we investigated them.

Modified Stitch Python backdoor

MD5 EC993FF561CBC175953502452BFA554A
Compilation date 2008-Nov-10 09:40:35
File type PE32 executable (DLL) (GUI) Intel 80386, for MS Windows
File size 7259 KB
File names flashplayer32_xa_pp_install.exe
flashplayer32pp_xa_install.exe
C2 server system0_update04driver_roots.dynamic-dns[.]net:443

This implant is a modified version of the open-source Python backdoor called Stitch, packed as a standalone PE executable with Py2exe.

Threat actors wrapped Stitch with custom Python code to perform additional operations:

  • It downloads a legitimate Adobe Flash installation program from the C2 server at startup;
  • It auto-updates the backdoor from ubntrooters.serveuser[.]com at startup;
  • It ensures persistence through schtasks [T1053] with a logon task named AdobeUpdater pointing to C:ProgramDatapackageAdobeService.exe.

Under the hood, Stitch is a remote shell program that provides classic backdoor functionalities by establishing a direct socket connection, to exchange AES-encrypted data with the remote server.

Conclusion

With almost 10 compromised websites and dozens of implanted hosts (that we know of), the attackers have set up a sizable yet very targeted water-holing attack. The toolset that’s being used seems low-budget and not fully developed, but has been modified several times in a few months to leverage interesting features like Google Drive C2, and would be characteristic of a small, agile team.

We were unable to observe any live operations, but some tracks indicate that the Godlike12 backdoor is not widespread, and is probably used to conduct reconnaissance and data-exfiltration operations.

We were unable to correlate these attacks to any known APT groups.
For more details and the latest information on this threat actor, please contact [email protected]

Appendix – IOCs

Infrastructure

Domain IP address Description
root20system20macosxdriver.serveusers[.]com 45.32.154[.]111 Watering hole targets validator server
loginwebmailnic.dynssl[.]com 207.148.117[.]159 Watering hole targets validator server
ubntrooters.serveuser[.]com 45.76.43[.]153 Stitch auto-update server
system0_update04driver_roots.dynamic-dns[.]net 95.179.171[.]173 Stitch C2
sys_andriod20_designer.dynamic-dns[.]net 45.63.114[.]152 Stitch C2
adobeflash31_install.ddns[.]info 95.179.171[.]173 Installer package C2
airjaldinet[.]ml 108.61.178[.]125 Older C++ validator C2

URLs

https://loginwebmailnic.dynssl[.]com/part/mac/contentmc.php
https://loginwebmailnic.dynssl[.]com/all/content.php
https://loginwebmailnic.dynssl[.]com/lh/content.php
https://root20system20macosxdriver.serveusers[.]com/yW6jOyQM16rj.html
https://root20system20macosxdriver.serveusers[.]com/itV6E1uKYiOo.html
http://ubntrooters.serveuser[.]com/wuservice.exe
http://ubntrooters.serveuser[.]com/upgrade.exe
http://ubntrooters.serveuser[.]com/flashplayer_update.exe
http://adobeflash31_install.ddns[.]info/flash/sys.txt
https://github[.]com/AdobeFlash32/FlashUpdate/
https://airjaldinet[.]ml/

Hashes (MD5)

0C6025A2C68E1C702A3022F1A6AE9169
1076A0EE924F198A7BD58A2DE1F060A0
10B4D3A667E06DC4B06AA542173D052C
11294E27491B496E36CA7DB9F363ADCD
11A16E109DBAF2FD080D8490328DE5A1
2E1862BC23085402EE11C88E540533C0
3989AC9EFB6A725918BD1810765D30B3
481DD1A37C86FDA68BCED0ECB2F47597
5287045D15FF60618F426AFC03BBB331
53CB974CAF909EEDCD86D2F80E75AD0A
5F19BB1688CA836B9207248F9096B9D2
6DF39D2CE9FCA27B78CC5CA0BED89703
7EB0C103AE21189AD9AD4A9804293B22
8623FA35226AC92CF6F02447AC80AFB0
9E69DDE252038B4A38EF0BFF6CE7FCD7
AD7A4333BC364DF3D4FA00B13CBBBEB4
B02ABA86409BE2AB263B1A476C1A1417
B21AF331B1752A70360B5D8DC9013F3F
B21BD93F15916A9A4AC76350D8FDBE10
BE3E563E95DEDCA0CEC9792194FFF2AC
DE2D8AF2EFED0C145690B2F13CD063B3
EC993FF561CBC175953502452BFA554A
ED081A869D30BB90B76552C83BD784C8
BEC4482890A89F0184B463C727709D53
9A819F2CE060058745FF5374221ADA7C
6DC5F8282DF76F4045F75FEA3277DF41


Temp Mails (https://tempemail.co/) is a new free temporary email addresses service. This service provide you random 10 minutes emails addresses. It is also known by names like: temporary mail, disposable mail, throwaway email, one time mail, anonymous email address… All emails received by Tempmail servers are displayed automatically in your online browser inbox.

iOS exploit chain deploys “LightSpy” feature-rich malware – 10 minute mail

A watering hole was discovered on January 10, 2020 utilizing a full remote iOS exploit chain to deploy a feature-rich implant named LightSpy. The site appears to have been designed to target users in Hong Kong based on the content of the landing page. Since the initial activity, we released two private reports exhaustively detailing spread, exploits, infrastructure and LightSpy implants.

Landing page of watering hole site

We are temporarily calling this APT group “TwoSail Junk”. Currently, we have hints from known backdoor callbacks to infrastructure about clustering this campaign with previous activity. And we are working with colleagues to tie LightSpy with prior activity from a long running Chinese-speaking APT group, previously reported on as Spring Dragon/Lotus Blossom/Billbug(Thrip), known for their Lotus Elise and Evora backdoor malware. Considering this LightSpy activity has been disclosed publicly by our colleagues from TrendMicro, we would like to further contribute missing information to the story without duplicating content. And, in our quest to secure technologies for a better future, we reported the malware and activity to Apple and other relevant companies.

This supplemental information can be difficult to organize to make for easy reading. In light of this, this document is broken down into several sections.

  1. Deployment timeline – additional information clarifying LightSpy deployment milestone events, including both exploit releases and individual LightSpy iOS implant component updates.
  2. Spreading – supplemental technical details on various techniques used to deliver malicious links to targets
  3. Infrastructure – supplemental description of a TwoSail Junk RDP server, the LightSpy admin panel, and some related server-side javascript
  4. Android implant and a pivot into evora – additional information on an Android implant and related infrastructure. After pivoting from the infrastructure in the previous section, we find related implants and backdoor malware, helping to connect this activity to previously known SpringDragon APT with low confidence.

More information about LightSpy is available to customers of Kaspersky Intelligence Reporting. Contact: [email protected]

Deployment timeline

During our investigation, we observed the actor modifying some components involved in the exploit chain on February 7, 2020 with major changes, and on March 5, 2020 with minor ones.

Figure 1. Brief LightSpy event timeline

The first observed version of the WebKit exploit dated January 10, 2020 closely resembled a proof of concept (PoC), containing elements such as buttons, alert messages, and many log statements throughout. The second version commented out or removed many of the log statements, changed alert() to print() statements, and also introduced some language errors such as “your device is not support…” and “stab not find…”.

By analyzing the changes in the first stage WebKit exploit, we discovered the list of supported devices was also significantly extended:
Table 1. iOS version exploit support expansion

Device iOS version Supported as of Jan 10 Supported as of Feb 7
iPhone 6 11.03 +
iPhone 6S 12.01 + commented
12.2 +
iPhone 7 12.1 +
12.11 + +
12.12 + +
12.14 +
12.2 +
iPhone 7+ 12.2 +
iPhone 8 12.2 +
iPhone 8+ 12.2 +
iPhone X 12.2 +

As seen above, the actor was actively changing implant components, which is why we are providing a full list of historical hashes in the IoC section at the end of this report. There were many minor changes that did not directly affect the functionality of each component, but there were also some exceptions to this that will be expanded on below. Based on our observations of these changes over a relatively short time frame, we can assess that the actor implemented a fairly agile development process, with time seemingly more important than stealthiness or quality.

One interesting observation involved the “EnvironmentalRecording” plugin (MD5: ae439a31b8c5487840f9ad530c5db391), which was a dynamically linked shared library responsible for recording surrounding audio and phone calls. On February 7, 2020, we noticed a new binary (MD5: f70d6b3b44d855c2fb7c662c5334d1d5) with the same name with no similarities to the earlier one. This new file did not contain any environment paths, version stamps, or any other traces from the parent plugin pattern. Its sole purpose was to clean up the implant components by erasing all files located in “/var/iolight/”, “/bin/light/”, and “/bin/irc_loader/”. We’re currently unsure whether the actor intended to replace the original plugin with an uninstall package or if this was a result of carelessness or confusion from the rapid development process.

Another example of a possible mistake involved the “Screenaaa” plugin. The first version (MD5: 35fd8a6eac382bfc95071d56d4086945) that was deployed on January 10, 2020 did what we expected: It was a small plugin designed to capture a screenshot, create a directory, and save the capture file in JPEG format. However, the plugin (MD5: 7b69a20920d3b0e6f0bffeefdce7aa6c) with the same name that was packaged on February 7 had a completely different functionality. This binary was actually a LAN scanner based on MMLanScan, an open source project for iOS that helps scan a network to show available devices along with their MAC addresses, hostname, and manufacturer. Most likely, this plugin was mistakenly bundled up in the February 7 payload with the same name as the screenshot plugin.

Figure 2. LightSpy iOS implant component layout and communications

Spreading

We cannot say definitively that we have visibility into all of their spreading mechanisms. We do know that in past campaigns, precise targeting of individuals was performed over various social network platforms with direct messaging. And, both ours and previous reporting from others have documented TwoSail Junk’s less precise and broad use of forum posts and replies. These forum posts direct individuals frequenting these sites to pages hosting iframes served from their exploit servers. We add Telegram channels and instagram posts to the list of communication channels abused by these attackers.

These sites and communication medium are known to be frequented by some activist groups.

Figure 3. LightSpy iPhone infection steps

The initial watering hole site (hxxps://appledaily.googlephoto[.]vip/news[.]html) on January 10, 2020 was designed to mimic a well known Hong Kong based newspaper “Apple Daily” by copy-pasting HTML content from the original:

Figure 4. Source of html page mimicking newspaper “Apple Daily”

However, at that time, we had not observed any indications of the site being purposely distributed in the wild. Based on our KSN detection statistics, we began seeing a massive distribution campaign beginning on February 18, 2020.

Table 2. LightSpy related iframe domains, urls, and first seen timestamps

Starting on February 18, the actors began utilizing a series of invisible iframes to redirect potential victims to the exploit site as well as the intended legitimate news site from the lure.

Figure 5. Source of html page with lure and exploit

Infrastructure

RDP Clues

The domain used for the initial watering hole page (googlephoto[.]vip) was registered through GoDaddy on September 24, 2019. No unmasked registration information was able to be obtained for this domain. The subdomain (appledaily.googlephoto[.]vip) began resolving to a non-parked IP address (103.19.9[.]185) on January 10, 2020 and has not moved since. The server is located in Singapore and is hosted by Beyotta Network, LLP.

At the time of our initial investigation, the server was listening on ports 80 (HTTP) and 3389 (RDP with SSL/TLS enabled). The certificate for the server was self-signed and created on December 16, 2019. Based on Shodan data as early as December 21, 2019, there was a currently logged in user detected who’s name was “SeinandColt”.

Figure 6. Screenshot of RDP login page for the server 103.19.9[.]185

Admin Panel

The C2 server for the iOS payload (45.134.1[.]180) also appeared to have an admin panel on TCP port 50001.

The admin panel seems to be a Vue.js application bundled with Webpack. It contains two language packs: English and Chinese. A cursory analysis provides us the impression of actual scale of the framework:

If we take a closer look at the index.js file for the panel, some interesting configurations are visible, to include a user config, an application list, log list, and other interesting settings.

The “userConfig” variable indicates other possible platforms that may have been targeted by the same actors, such as linux, windows, and routers.

Another interesting setting includes the “app_list” variable which is commented out. This lists two common applications used for streaming and chat mostly in China (QQ and Miapoi). Looking further, we can also see that the default map coordinates in the config point directly to the Tian’anmen Gate in Beijing, however, most likely this is just a common and symbolic mapping application default for the center of Beijing.

Android implants and a pivot into “evora”

During analysis of the infrastructure related to iOS implant distribution we also found a link directing to Android malware – hxxp://app.hkrevolution[.]club/HKcalander[.]apk (MD5: 77ebb4207835c4f5c4d5dfe8ac4c764d).

According to artefacts found in google cache, this link was distributed through Telegram channels “winuxhk” and “brothersisterfacebookclub”, and Instagram posts in late November 2019 with a message lure in Chinese translated as “The Hong Kong People Calendar APP is online ~~~ Follow the latest Hong Kong Democracy and Freedom Movement. Click to download and support the frontline. Currently only Android version is available.”

Further technical analysis of the packed APK reveals the timestamp of its actual build – 2019-11-04 18:12:33. Also it uses the subdomain, sharing an iOS implant distribution domain, as its c2 server – hxxp://svr.hkrevolution[.]club:8002.

Its code contains a link to another related domain:

Checking this server we found it hosted another related APK:

MD5 fadff5b601f6fca588007660934129eb
URL hxxp://movie.poorgoddaay[.]com/MovieCal[.]apk
C2 hxxp://app.poorgoddaay[.]com:8002
Build timestamp 2019-07-25 21:57:47

The distribution vector remains the same – Telegram channels:

The latest observed APK sample is hosted on a server that is unusual for the campaign context – xxinc-media[.]oss-cn-shenzhen.aliyuncs[.]com. We assume that the actors are taking steps to split the iOS and Android activities between different infrastructure pieces.

MD5 5d2b65790b305c186ef7590e5a1f2d6b
URL hxxps://xxinc-media.oss-cn-shenzhen.aliyuncs[.]com/calendar-release-1.0.1.apk
C2 hxxp://45.134.0[.]123:8002
Build timestamp 2020-01-14 18:30:30

We had not observed any indications of this URL being distributed in the wild yet.

If we take a look closer at the domain poorgoddaay[.]com that not only hosted the malicious APK but also was a C2 for them, we can note that there are two subzones of particular interest to us:

  • zg.poorgoddaay[.]com
  • ns1.poorgoddaay[.]com

We were able to work with partners to pivot into a handful of “evora” samples that use the above two subzones as their C2. Taking that a step further, using our Kaspersky Threat Attribution Engine (KTAE), we can see that the partner samples using those subzones are 99% similar to previous backdoors deployed by SpringDragon.

We are aware of other related and recent “evora” malware samples calling back to these same subnets while targeting organizations in Hong Kong as well. These additional factors help lend at least low confidence to clustering this activity with SpringDragon/LotusBlossom/Billbug.

Conclusion

This particular framework and infrastructure is an interesting example of an agile approach to developing and deploying surveillance framework in Southeast Asia. This innovative approach is something we have seen before from SpringDragon, and LightSpy targeting geolocation at least falls within previous regional targeting of SpringDragon/LotusBlossom/Billbug APT, as does infrastructure and “evora” backdoor use.

Indicators of Compromise

File hashes

payload.dylib
9b248d91d2e1d1b9cd45eb28d8adff71 (Jan 10, 2020)
4fe3ca4a2526088721c5bdf96ae636f4 (Feb 7, 2020)

ircbin.plist
e48c1c6fb1aa6c3ff6720e336c62b278 (Jan 10, 2020)

irc_loader
53acd56ca69a04e13e32f7787a021bb5 (Jan 10, 2020)

light
184fbbdb8111d76d3b1377b2768599c9 (Jan 10, 2020)
bfa6bc2cf28065cfea711154a3204483 (Feb 7, 2020)
ff0f66b7089e06702ffaae6025b227f0 (Mar 5, 2020)

baseinfoaaa.dylib
a981a42fb740d05346d1b32ce3d2fd53 (Jan 10, 2020)
5c69082bd522f91955a6274ba0cf10b2 (Feb 7, 2020)

browser
7b263f1649dd56994a3da03799611950 (Jan 10, 2020)

EnvironmentalRecording
ae439a31b8c5487840f9ad530c5db391 (Jan 10, 2020)
f70d6b3b44d855c2fb7c662c5334d1d5 (Feb 7, 2020)

FileManage
f1c899e7dd1f721265cc3e3b172c7e90 (Jan 10, 2020)
ea9295d8409ea0f1d894d99fe302070e (Feb 7, 2020)

ios_qq
c450e53a122c899ba451838ee5250ea5 (Jan 10, 2020)
f761560ace765913695ffc04dfb36ca7 (Feb 7, 2020)

ios_telegram
1e12e9756b344293352c112ba84533ea (Jan 10, 2020)
5e295307e4429353e78e70c9a0529d7d (Feb 7, 2020)

ios_wechat
187a4c343ff4eebd8a3382317cfe5a95 (Jan 10, 2020)
66d2379318ce8f74cfbd0fb26afc2084 (Feb 7, 2020)

KeyChain
db202531c6439012c681328c3f8df60c (Jan 10, 2020)

locationaaa.dylib
3e7094eec0e99b17c5c531d16450cfda (Jan 10, 2020)
06ff47c8108f7557bb8f195d7b910882 (Feb 7, 2020)

Screenaaa
35fd8a6eac382bfc95071d56d4086945 (Jan 10, 2020)
7b69a20920d3b0e6f0bffeefdce7aa6c (Feb 7, 2020)

ShellCommandaaa
a8b0c99f20a303ee410e460730959d4e (Jan 10, 2020)

SoftInfoaaa
8cdf29e9c6cca6bf8f02690d8c733c7b (Jan 10, 2020)

WifiList
c400d41dd1d3aaca651734d4d565997c (Jan 10, 2020)

Android malware
77ebb4207835c4f5c4d5dfe8ac4c764d
fadff5b601f6fca588007660934129eb
5d2b65790b305c186ef7590e5a1f2d6b

Past similar SpringDragon evora
1126f8af2249406820c78626a64d12bb
33782e5ba9067b38d42f7ecb8f2acdc8

Domains and IPs

Implant c2
45.134.1[.]180 (iOS)
45.134.0[.]123 (Android)
app.poorgoddaay[.]com (Android)
svr[.]hkrevolution[.]club (Android)

WebKit exploit landing
45.83.237[.]13
messager[.]cloud

Spreading
appledaily.googlephoto[.]vip
www[.]googlephoto[.]vip
news2.hkrevolution[.]club
news.hkrevolution[.]club
www[.]facebooktoday[.]cc
www[.]hkrevolt[.]com
news.hkrevolt[.]com
movie.poorgoddaay[.]com
xxinc-media[.]oss-cn-shenzhen.aliyuncs[.]com

Related subdomains
app.hkrevolution[.]club
news.poorgoddaay[.]com
zg.poorgoddaay[.]com
ns1.poorgoddaay[.]com

Full Mobile Device Command List

change_config
exe_cmd
stop_cmd
get_phoneinfo
get_contacts
get_call_history
get_sms
delete_sms
send_sms
get_wechat_account
get_wechat_contacts
get_wechat_group
get_wechat_msg
get_wechat_file
get_location
get_location_coninuing
get_browser_history
get_dir
upload_file
download_file
delete_file
get_picture
get_video
get_audio
create_dir
rename_file
move_file
copy_file
get_app
get_process
get_wifi_history
get_wifi_nearby
call_record
call_photo
get_qq_account
get_qq_contacts
get_qq_group
get_qq_msg
get_qq_file
get_keychain
screenshot


Temp Mails (https://tempemail.co/) is a new free temporary email addresses service. This service provide you random 10 minutes emails addresses. It is also known by names like: temporary mail, disposable mail, throwaway email, one time mail, anonymous email address… All emails received by Tempmail servers are displayed automatically in your online browser inbox.

WildPressure targets industrial-related entities in the Middle East – 10 minute mail

In August 2019, Kaspersky discovered a malicious campaign distributing a fully fledged C++ Trojan that we call Milum. All the victims we registered were organizations from the Middle East. At least some of them are related to industrial sector. Our Kaspersky Threat Attribution Engine (KTAE) doesn’t show any code similarities with known campaigns. Nor have we seen any target intersections. In fact, we found just three almost unique samples, all in one country. So we consider the attacks to be targeted and have currently named this operation WildPressure.

The compilation timestamps for all these files is the same – March 2019. This is coherent with the fact that we registered no infections before May 31, 2019, so the compilation dates don’t seem to be spoofed. For their campaign infrastructure the operators used rented OVH and Netzbetrieb virtual private servers (VPS) and a domain registered with the Domains by Proxy anonymization service.

The malware uses the JSON format for configuration data and as a C2 communication protocol over HTTP as well. Inside the encrypted communications within the HTTP POST requests we found several interesting fields. One of them shows the malware version – 1.0.1. A version number like this indicates an early stage of development. Other fields suggest the existence of, at the very least, plans for non-C++ versions.

The only encryption implemented is the RC4 algorithm with different 64-byte keys for different victims. Also, the developers were kind enough to leave RTTI data inside the files. Kaspersky products detect this malware as Backdoor.Win32.Agent. For more information, please contact: [email protected]

Why we call it Milum and why it’s of interest

All the aforementioned C++ Trojans are compiled as standalone PE files, originally named Milum46_Win32.exe. The word ‘milum’ is used in the C++ class names inside the malware, so we named the Trojan after it.

Another distinctive characteristic is that the malware exports lots of zlib compression functions, such as zlibVersion(), inflate() or deflate(). This compression is needed for C2 communication, but in reality there is no need to export them in the case of a standalone application.

The JSON configuration fields are not limited to just the version and programming language; the campaign operators also use target IDs that are found in the samples. Among them, we found HatLandM30 and HatLandid3 – neither of which we are familiar with. The following table provides Milum samples that have similar PE header compilation timestamps but different target IDs:

Milum46_Win32.exe sample MD5 hash Timestamp (GMT) clientid
0C5B15D89FDA9BAF446B286C6F97F535 2019.03.09 06:17:19 839ttttttt
17B1A05FC367E52AADA7BDE07714666B 2019.03.09 06:17:19 HatLandid3
A76991F15D6B4F43FBA419ECA1A8E741 2019.03.09 06:17:19 HatLandM30

Rather than describing all the configuration fields one by one, we have gathered them together in the following table, with all the main characteristics for this malware family:

Programming language C++ with STL functions used mostly to parse JSON data and exception handling.
Configuration data Base64-encoded JSON data in PE resources. Includes timeouts, C2 URLs and keys for communication, including RC4 64-byte key.
Network protocol Trojan transmits compressed JSON data in HTTP POST requests with gzip, base64-encoded and RC4 encrypted.
Beacon data Encrypted JSON contains the malware version “1.0.1”, Epoch timestamp and client id. It also has specific fields such as “vt” and “ext” that correspond to programming language “c++” and file extension “exe”. If our hypothesis is correct, this suggests that non-C++ Trojan versions may be planned, if not already implemented.
Persistence HKCU autorun system registry keys Run and RunOnce.
Encryption The communication encryption used is RC4 with the 64-byte key stored in the configuration data.
Compression For compression the Trojan uses an embedded gzip code. For some reason gzip functions are exported from PE, although the samples are standalone executables, not DLLs.

Let’s dig a little deeper inside

The most popular sample in our telemetry was:

SHA256 a1ad9301542cc23a04a57e6567da30a6e14eb24bf06ce9dd945bbadf17e4cf56
MD5    0c5b15d89fda9baf446b286c6f97f535
Compiled     2019.03.09 06:17:19 (GMT)
Size   520704
Internal name       Milum46_Win32.exe

This application exists as an invisible toolbar window. The main malicious functions are implemented in a separate thread. Milum decodes its configuration data and, besides timeouts, it gets the parameters “clientid” and “encrypt_key” to use in RC4 encryption.

Example of the decoded and beautified configuration data. The “clientid” field differs in every sample observed

The following table describes the different configuration parameters:

Config parameter Parameter features
shortwait Pause in milliseconds between C2 communication working cycles
clientid Unique ASCII target name
encrypt_key RC4 encryption key for JSON-based C2 communications
relays – url Full URL to send HTTP POST beacon and GET commands
relays – key Unique ASCII key for each C2 to communicate with it

The operators can run the Trojan using the key (“b” or “B”) as the first argument and the file name as the second. In this case, Milum will delete the file sent as a parameter. Then the Trojan will create the C:ProgramDataMicappWindows directory and parse its configuration data to form the beacon to send to its C2.

To send the beacon, Milum uses the HTTP POST request with three parameters as enumerated in the table below.

Beacon parameter Parameter values
md Clientid from config, with prefix 01011 and random five-character ASCII suffix
nk Key from config to communicate with C2, differs for each server
val Compressed, encrypted and encoded command JSON data

The first two parameters are taken from the configuration data. The third one is encrypted and after decryption, decompression, decoding and beautifying, it looks like this:

Decoded and beautified JSON beacon to C2. In this case, the connection to the first server was unsuccessful

There are several fields worth mentioning here. We referred above to different programming languages besides C++: “vt” seems to reference a programming language and “ext” a file extension. The only reason that we could think of for keeping these is if the attackers have several Trojans, written in different languages, to work with the same control server.

Regarding the “command” field, the control servers were inaccessible at the time of the analysis, so we don’t have commands from them. However, we analyzed the command handlers in Milum’s code as described below:

Code Meaning Features
1 Execution Silently execute received interpreter command and return result through pipe
2 Server to client Decode received content in “data” JSON field and drop to file mentioned in “path” field
3 Client to server Encode file mentioned in received command “path” field to send it
4 File info Get file attributes: hidden, read only, archive, system or executable
5 Cleanup Generate and run batch script to delete itself
6 Command result Get command execution status
7 System information Validate target with Windows version, architecture (32- or 64-bit), host and user name, installed security products (with WQL request “Select From AntiVirusProduct WHERE displayName <>’Windows Defender’”)
8 Directory list Get info about files in directory: hidden, read only, archive, system or executable
9 Update Get the new version and remove the old one

Who was attacked?

According to our telemetry, the Milum Trojan was exclusively used to attack targets in the Middle East from at least the end of May 2019.

Number of detections for one of the samples from September 2019

We were able to sinkhole one of the WildPressure C2 domains (upiserversys1212[.]com) in September 2019. The vast majority of visitor IPs were also from the Middle East, and we believe the rest were network scanners, TOR exit nodes or VPN connections.

C2 domain sinkholing also shows active infections mostly from the Middle East

And who’s behind it?

To date we haven’t observed any strong code- or victim-based similarities with any known actor or set of activity. Their C++ code is quite common, regarding configuration data and communication protocol malware uses base64-encoded JSON-formatted configuration data stored in the binary’s resource section and parses it with Standard Template Library (STL) functions. However, these commonalities are not conclusive enough for attribution and our hypothesis is that they are merely coincidence. We would continue to monitoring this activity

To sum up

To date, we don’t have any data regarding Milum’s spreading mechanism. A campaign that is, apparently, exclusively targeting entities in the Middle East (at least some of them are industrial-related) is something that automatically attracts the attention of any analyst. Any similarities should be considered weak in terms of attribution, and are may simply be techniques copied from previous well-known cases. Indeed, this “learning from more experienced attackers” cycle has been adopted by some new interesting actors in recent years.

We should also be cautious regarding the true targeting of this new set of activities, as it is probably too soon to jump to conclusions. The targeted nature seems to be clear, but the targeting itself might be limited by our own visibility. The malware is not exclusively designed against any kind of victim in particular and might be reused in other operations.

Indicators of compromise

Files MD5
0C5B15D89FDA9BAF446B286C6F97F535
17B1A05FC367E52AADA7BDE07714666B
A76991F15D6B4F43FBA419ECA1A8E741
Original file names are Milum46_Win32.exe; on the target side they exist as system32.exe

URLs
upiserversys1212[.]com/rl.php
37.59.87[.]172/page/view.php
80.255.3[.]86/page/view.php


Temp Mails (https://tempemail.co/) is a new free temporary email addresses service. This service provide you random 10 minutes emails addresses. It is also known by names like: temporary mail, disposable mail, throwaway email, one time mail, anonymous email address… All emails received by Tempmail servers are displayed automatically in your online browser inbox.

WildPressure targets industrial-related entities in the Middle East – 10 minute mail

In August 2019, Kaspersky discovered a malicious campaign distributing a fully fledged C++ Trojan that we call Milum. All the victims we registered were organizations from the Middle East. At least some of them are related to industrial sector. Our Kaspersky Threat Attribution Engine (KTAE) doesn’t show any code similarities with known campaigns. Nor have we seen any target intersections. In fact, we found just three almost unique samples, all in one country. So we consider the attacks to be targeted and have currently named this operation WildPressure.

The compilation timestamps for all these files is the same – March 2019. This is coherent with the fact that we registered no infections before May 31, 2019, so the compilation dates don’t seem to be spoofed. For their campaign infrastructure the operators used rented OVH and Netzbetrieb virtual private servers (VPS) and a domain registered with the Domains by Proxy anonymization service.

The malware uses the JSON format for configuration data and as a C2 communication protocol over HTTP as well. Inside the encrypted communications within the HTTP POST requests we found several interesting fields. One of them shows the malware version – 1.0.1. A version number like this indicates an early stage of development. Other fields suggest the existence of, at the very least, plans for non-C++ versions.

The only encryption implemented is the RC4 algorithm with different 64-byte keys for different victims. Also, the developers were kind enough to leave RTTI data inside the files. Kaspersky products detect this malware as Backdoor.Win32.Agent. For more information, please contact: [email protected]

Why we call it Milum and why it’s of interest

All the aforementioned C++ Trojans are compiled as standalone PE files, originally named Milum46_Win32.exe. The word ‘milum’ is used in the C++ class names inside the malware, so we named the Trojan after it.

Another distinctive characteristic is that the malware exports lots of zlib compression functions, such as zlibVersion(), inflate() or deflate(). This compression is needed for C2 communication, but in reality there is no need to export them in the case of a standalone application.

The JSON configuration fields are not limited to just the version and programming language; the campaign operators also use target IDs that are found in the samples. Among them, we found HatLandM30 and HatLandid3 – neither of which we are familiar with. The following table provides Milum samples that have similar PE header compilation timestamps but different target IDs:

Milum46_Win32.exe sample MD5 hash Timestamp (GMT) clientid
0C5B15D89FDA9BAF446B286C6F97F535 2019.03.09 06:17:19 839ttttttt
17B1A05FC367E52AADA7BDE07714666B 2019.03.09 06:17:19 HatLandid3
A76991F15D6B4F43FBA419ECA1A8E741 2019.03.09 06:17:19 HatLandM30

Rather than describing all the configuration fields one by one, we have gathered them together in the following table, with all the main characteristics for this malware family:

Programming language C++ with STL functions used mostly to parse JSON data and exception handling.
Configuration data Base64-encoded JSON data in PE resources. Includes timeouts, C2 URLs and keys for communication, including RC4 64-byte key.
Network protocol Trojan transmits compressed JSON data in HTTP POST requests with gzip, base64-encoded and RC4 encrypted.
Beacon data Encrypted JSON contains the malware version “1.0.1”, Epoch timestamp and client id. It also has specific fields such as “vt” and “ext” that correspond to programming language “c++” and file extension “exe”. If our hypothesis is correct, this suggests that non-C++ Trojan versions may be planned, if not already implemented.
Persistence HKCU autorun system registry keys Run and RunOnce.
Encryption The communication encryption used is RC4 with the 64-byte key stored in the configuration data.
Compression For compression the Trojan uses an embedded gzip code. For some reason gzip functions are exported from PE, although the samples are standalone executables, not DLLs.

Let’s dig a little deeper inside

The most popular sample in our telemetry was:

SHA256 a1ad9301542cc23a04a57e6567da30a6e14eb24bf06ce9dd945bbadf17e4cf56
MD5    0c5b15d89fda9baf446b286c6f97f535
Compiled     2019.03.09 06:17:19 (GMT)
Size   520704
Internal name       Milum46_Win32.exe

This application exists as an invisible toolbar window. The main malicious functions are implemented in a separate thread. Milum decodes its configuration data and, besides timeouts, it gets the parameters “clientid” and “encrypt_key” to use in RC4 encryption.

Example of the decoded and beautified configuration data. The “clientid” field differs in every sample observed

The following table describes the different configuration parameters:

Config parameter Parameter features
shortwait Pause in milliseconds between C2 communication working cycles
clientid Unique ASCII target name
encrypt_key RC4 encryption key for JSON-based C2 communications
relays – url Full URL to send HTTP POST beacon and GET commands
relays – key Unique ASCII key for each C2 to communicate with it

The operators can run the Trojan using the key (“b” or “B”) as the first argument and the file name as the second. In this case, Milum will delete the file sent as a parameter. Then the Trojan will create the C:ProgramDataMicappWindows directory and parse its configuration data to form the beacon to send to its C2.

To send the beacon, Milum uses the HTTP POST request with three parameters as enumerated in the table below.

Beacon parameter Parameter values
md Clientid from config, with prefix 01011 and random five-character ASCII suffix
nk Key from config to communicate with C2, differs for each server
val Compressed, encrypted and encoded command JSON data

The first two parameters are taken from the configuration data. The third one is encrypted and after decryption, decompression, decoding and beautifying, it looks like this:

Decoded and beautified JSON beacon to C2. In this case, the connection to the first server was unsuccessful

There are several fields worth mentioning here. We referred above to different programming languages besides C++: “vt” seems to reference a programming language and “ext” a file extension. The only reason that we could think of for keeping these is if the attackers have several Trojans, written in different languages, to work with the same control server.

Regarding the “command” field, the control servers were inaccessible at the time of the analysis, so we don’t have commands from them. However, we analyzed the command handlers in Milum’s code as described below:

Code Meaning Features
1 Execution Silently execute received interpreter command and return result through pipe
2 Server to client Decode received content in “data” JSON field and drop to file mentioned in “path” field
3 Client to server Encode file mentioned in received command “path” field to send it
4 File info Get file attributes: hidden, read only, archive, system or executable
5 Cleanup Generate and run batch script to delete itself
6 Command result Get command execution status
7 System information Validate target with Windows version, architecture (32- or 64-bit), host and user name, installed security products (with WQL request “Select From AntiVirusProduct WHERE displayName <>’Windows Defender’”)
8 Directory list Get info about files in directory: hidden, read only, archive, system or executable
9 Update Get the new version and remove the old one

Who was attacked?

According to our telemetry, the Milum Trojan was exclusively used to attack targets in the Middle East from at least the end of May 2019.

Number of detections for one of the samples from September 2019

We were able to sinkhole one of the WildPressure C2 domains (upiserversys1212[.]com) in September 2019. The vast majority of visitor IPs were also from the Middle East, and we believe the rest were network scanners, TOR exit nodes or VPN connections.

C2 domain sinkholing also shows active infections mostly from the Middle East

And who’s behind it?

To date we haven’t observed any strong code- or victim-based similarities with any known actor or set of activity. Their C++ code is quite common, regarding configuration data and communication protocol malware uses base64-encoded JSON-formatted configuration data stored in the binary’s resource section and parses it with Standard Template Library (STL) functions. However, these commonalities are not conclusive enough for attribution and our hypothesis is that they are merely coincidence. We would continue to monitoring this activity

To sum up

To date, we don’t have any data regarding Milum’s spreading mechanism. A campaign that is, apparently, exclusively targeting entities in the Middle East (at least some of them are industrial-related) is something that automatically attracts the attention of any analyst. Any similarities should be considered weak in terms of attribution, and are may simply be techniques copied from previous well-known cases. Indeed, this “learning from more experienced attackers” cycle has been adopted by some new interesting actors in recent years.

We should also be cautious regarding the true targeting of this new set of activities, as it is probably too soon to jump to conclusions. The targeted nature seems to be clear, but the targeting itself might be limited by our own visibility. The malware is not exclusively designed against any kind of victim in particular and might be reused in other operations.

Indicators of compromise

Files MD5
0C5B15D89FDA9BAF446B286C6F97F535
17B1A05FC367E52AADA7BDE07714666B
A76991F15D6B4F43FBA419ECA1A8E741
Original file names are Milum46_Win32.exe; on the target side they exist as system32.exe

URLs
upiserversys1212[.]com/rl.php
37.59.87[.]172/page/view.php
80.255.3[.]86/page/view.php


Temp Mails (https://tempemail.co/) is a new free temporary email addresses service. This service provide you random 10 minutes emails addresses. It is also known by names like: temporary mail, disposable mail, throwaway email, one time mail, anonymous email address… All emails received by Tempmail servers are displayed automatically in your online browser inbox.

Operation AppleJeus Sequel | Securelist – 10 minute mail

The Lazarus group is currently one of the most active and prolific APT actors. In 2018, Kaspersky published a report on one of their campaigns, named Operation AppleJeus. Notably, this operation marked the first time Lazarus had targeted macOS users, with the group inventing a fake company in order to deliver their manipulated application and exploit the high level of trust among potential victims. As a result of our ongoing efforts, we identified significant changes to the group’s attack methodology. To attack macOS users, the Lazarus group has developed homemade macOS malware, and added an authentication mechanism to deliver the next stage payload very carefully, as well as loading the next-stage payload without touching the disk. In addition, to attack Windows users, they have elaborated a multi-stage infection procedure, and significantly changed the final payload. We assess that the Lazarus group has been more careful in its attacks following the release of Operation AppleJeus and they have employed a number of methods to avoid being detected.

For more information, please contact: [email protected]

Life after Operation AppleJeus

After releasing Operation AppleJeus, the Lazarus group continued to use a similar modus operandi in order to compromise cryptocurrency businesses. We found more macOS malware similar to that used in the original Operation AppleJeus case. This macOS malware used public source code in order to build crafted macOS installers. The malware authors used QtBitcoinTrader developed by Centrabit.

Original AppleJeus WbBot case MacInstaller case
DMG file hash 48ded52752de9f9b73c6bf9ae81cb429 3efeccfc6daf0bf99dcb36f247364052 c2ffbf7f2f98c73b98198b4937119a18
PKG file hash dab34d94ca08ba5b25edadfe67ae4607 cb56955b70c87767dee81e23503086c3 8b4c532f10603a8e199aa4281384764e
PKG file name CelasTradePro.pkg WbBot.pkg BitcoinTrader.pkg
Packaging time 2018-07-12 14:09:33 2018-11-05 6:11:38 2018-12-19 0:15:19
Malicious mach-o hash aeee54a81032a6321a39566f96c822f5 b63e8d4277b190e2e3f5236f07f89eee bb04d77bda3ae9c9c3b6347f7aef19ac
C2 server www.celasllc[.]com/checkupdate.php https://www.wb-bot[.]org/certpkg.php https://www.wb-bot[.]org/certpkg.php
XOR key Moz&Wie;#t/6T!2y 6E^uAVd-^yYkB-XG 6E^uAVd-^yYkB-XG
RC4 key [email protected]%Df324V$Yd SkQpTUT8QEY&Lg+BpB SkQpTUT8QEY&Lg+BpB
2nd payload path /var/zdiffsec /var/pkglibcert /var/pkglibcert
2nd payload argument bf6a0c760cc642 bf6a0c760cc642 bf6a0c760cc642

These three macOS installers use a similar post installer script in order to implant a mach-o payload, as well as using the same command-line argument when executing the fetched second-stage payload. However, they have started changing their macOS malware. We recognized a different type of macOS malware, MarkMakingBot.dmg (be37637d8f6c1fbe7f3ffc702afdfe1d), created on 2019-03-12. It doesn’t have an encryption/decryption routine for network communication. We speculate that this is an intermediate stage in significant changes to their macOS malware.

Change of Windows malware

During our ongoing tracking of this campaign, we found that one victim was compromised by Windows AppleJeus malware in March 2019. Unfortunately, we couldn’t identify the initial installer, but we established that the infection started from a malicious file named WFCUpdater.exe. At that time, the actor used a fake website: wfcwallet[.]com

Fig. 1 Binary infection procedure used in WFCWallet case

The actor used a multi-stage infection like before, but the method was different. The infection started from .NET malware, disguised as a WFC wallet updater (a9e960948fdac81579d3b752e49aceda). Upon execution, this .NET executable checks whether the command line argument is “/Embedding” or not. This malware is responsible for decrypting the WFC.cfg file in the same folder with a hardcoded 20-byte XOR key (82 d7 ae 9b 36 7d fc ee 41 65 8f fa 74 cd 2c 62 b7 59 f5 62). This mimics the wallet updater connected to the C2 addresses:

  • wfcwallet.com (resolved ip: 108.174.195.134)
  • www.chainfun365.com (resolved ip: 23.254.217.53)

After that, it carries out the malware operator’s commands in order to install the next stage permanent payload. The actor delivered two more files into the victim’s system folder: rasext.dll and msctfp.dat. They used the RasMan (Remote Access Connection Manager) Windows service to register the next payload with a persistence mechanism. After fundamental reconnaissance, the malware operator implanted the delivered payload by manually using the following commands:

  • cmd.exe /c dir rasext.dll
  • cmd.exe /c dir msctfp.dat
  • cmd.exe /c tasklist /svc | findstr RasMan
  • cmd.exe /c reg add HKEY_LOCAL_MACHINESYSTEMCurrentControlSetservicesRasManThirdParty /v DllName /d rasext.dll /f

In order to establish remote tunneling, the actor delivered more tools, executing with command-line parameters. Unfortunately, we have had no chance to obtain this file, but we speculate that Device.exe is responsible for opening port 6378, and the CenterUpdater.exe tool was used for creating tunneling to a remote host. Note that the 104.168.167.16 server is used as a C2 server. The fake website hosting server for the UnionCryptoTrader case will be described next.

%APPDATA%LenovodevicecenterDevice.exe 6378

%APPDATA%LenovodevicecenterCenterUpdater.exe 127.0.0.1 6378 104.168.167.16 443

Change of macOS malware

JMTTrading case

While tracking this campaign, we identified more heavily deformed macOS malware. At the time, the attacker called their fake website and application JMTTrading. Other researchers and security vendors found it too, and published IoCs with abundant technical details. Malware Hunter Team tweeted about this malicious application, Vitali Kremez published a blog about the Windows version of the malware, and Object-See published details about the macOS malware. We believe these reports are sufficient to understand the technical side. Here, we would like to highlight what’s different about this attack.

  • The actor used GitHub in order to host their malicious applications.
  • The malware author used Object-C instead of QT framework in their macOS malware.
  • The malware implemented a simple backdoor function in macOS executable.
  • The malware encrypted/decrypted with a 16-byte XOR key (X,%`PMk–Jj8s+6=) similar to the previous case.
  • The Windows version of the malware used ADVobfuscator, a compiled time obfuscator, in order to hide its code.
  • The post-install script of macOS malware differed significantly from the previous version.

UnionCryptoTrader case

We also identified another macOS targeted attack that took place very recently. The malicious application name in this case is UnionCryptoTrader. After compiling a threat intelligence report for our customers, one security researcher (@dineshdina04) discovered an identical case, and Objective-See published a very detailed blog on the macOS malware used in this attack. The Objective-See blog goes into sufficient detail to explain the malware’s functionality, so we will just summarize the attack:

  • The post-install script is identical to that used in the JMTTrading case.
  • The malware author used SWIFT to develop this macOS malware.
  • The malware author changed the method for collecting information from the infected system.
  • The malware starts to conduct authentication using auth_signature and auth_timestamp parameters in order to deliver the second-stage payload more carefully. The malware acquires the current system time and combines it with the “12GWAPCT1F0I1S14” hardcoded string, and produces an MD5 hash of the combined string. This hash is used as the value of the auth_signature parameter and the current time is used as the value of the auth_timestamp parameter. The malware operator can reproduce the auth_signature value based on the auth_timestamp at the C2 server side.
  • The malware loads the next stage payload without touching the disk.

Windows version of UnionCryptoTrader

We also found a Windows version of the UnionCryptoTrader (0f03ec3487578cef2398b5b732631fec). It was executed from the Telegram messenger download folder:

C:Users[user name]DownloadsTelegram DesktopUnionCryptoTraderSetup.exe

We also found the actor’s Telegram group on their fake website. Based on these, we assess with high confidence that the actor delivered the manipulated installer using the Telegram messenger. Unfortunately, we can’t get all the related files as some payloads were only executed in memory. However, we can reassemble the whole infection procedure based on our telemetry. The overall infection procedure was very similar to the WFCWallet case, but with an added injection procedure, and they only used the final backdoor payload instead of using a tunneling tool.

Fig. 2 Binary infection procedure

The UnionCryptoTrader Windows version has the following window showing a price chart for several cryptocurrency exchanges.

Fig. 3 Windows version of UnionCryptoTrader

The Windows version of UnionCryptoTrader updater (629b9de3e4b84b4a0aa605a3e9471b31) has similar functionality to the macOS version. According to the build path (Z:Loaderx64ReleaseWinloaderExe.pdb), the malware author called this malware a loader. Upon launch, the malware retrieves the victim’s basic system information, sending it in the following HTTP POST format, as is the case with the macOS malware.

If the response code from the C2 server is 200, the malware decrypts the payload and loads it in memory. Finally, the malware sends the act=done value and return code. The next stage payload (e1953fa319cc11c2f003ad0542bca822), downloaded from this loader, is similar to the .NET downloader in the WFCWallet case. This malware is responsible for decrypting the Adobe.icx file in the same folder. It injects the next payload into the Internet Explorer process, and the tainted iexplore.exe process carries out the attacker’s commands. The final payload (dd03c6eb62c9bf9adaf831f1d7adcbab) is implanted manually as in the WFCWallet case. This final payload was designed to run only on certain systems. It seems that the malware authors produced and delivered malware that only works on specific systems based on previously collected information. The malware checks the infected system’s information and compares it to a given value. It seems the actor wants to execute the final payload very carefully, and wants to evade detection by behavior-based detection solutions.

Fig. 4 Malware execution flow

This Windows malware loads the encrypted msctfp.dat file in a system folder, and loads each configuration value. Then it executes an additional command based on the contents of this file. When the malware communicates with the C2 server, it uses a POST request with several predefined headers.

For the initial communication, the malware first sends parameters:

  • cgu: 64bits hex value from configuration
  • aip: MD5 hash value from configuration
  • sv: hardcoded value(1)

If the response code from the C2 server is 200, the malware sends the next POST request with encrypted data and a random value. The malware operator probably used the random value to identify each victim and verify the POST request.

  • imp: Random generated value
  • dsh: XORed value of imp
  • hb_tp: XORed value(key: 0x67BF32) of imp
  • hb_dl: Encrypted data to send to C2 server
  • ct: hardcoded value(1)

Finally, the malware downloads the next stage payload, decrypting it and possibly executing it with the Print parameter. We speculate that the DLL type payload will be downloaded and call its Print export function for further infection. We can’t get hold of the final payload that’s executed in memory, but we believe its backdoor-type malware is ultimately used to control the infected victim.

Infrastructures

We found several fake websites that were still online when we were investigating their infrastructure. They created fake cryptocurrency-themed websites, but they were far from perfect and most of the links didn’t work.

Fig. 5 Website of cyptian.com

Fig. 6 Website of unioncrypto.vip

We found an identical Cyptian web template on the internet. We speculate that the actor used free web templates like this to build their fake websites. Moreover, there is a Telegram address(@cyptian) on the Cyptian website. As we mentioned previously, the actor delivered a manipulated application via Telegram messenger. This Telegram address was still alive when we investigated, but there were no more activities at that time. According to the chat log, the group was created on December 17, 2018 and some accounts had already been deleted.

Fig. 7 Telegram account

Conclusion

We were able to identify several victims in this Operation AppleJeus sequel. Victims were recorded in the UK, Poland, Russia and China. Moreover, we were able to confirm that several of the victims are linked to cryptocurrency business entities.

Fig. 8 Infection map

The actor altered their macOS and Windows malware considerably, adding an authentication mechanism in the macOS downloader and changing the macOS development framework. The binary infection procedure in the Windows system differed from the previous case. They also changed the final Windows payload significantly from the well-known Fallchill malware used in the previous attack. We believe the Lazarus group’s continuous attacks for financial gain are unlikely to stop anytime soon.

Fig. 9 Timeline of Operation AppleJeus

Since the initial appearance of Operation AppleJeus, we can see that over time the authors have changed their modus operandi considerably. We assume this kind of attack on cryptocurrency businesses will continue and become more sophisticated.

Appendix I – Indicators of Compromise

File Hashes (malicious documents, Trojans, emails, decoys)

macOS malware

  • c2ffbf7f2f98c73b98198b4937119a18 MacInstaller.dmg
  • 8b4c532f10603a8e199aa4281384764e BitcoinTrader.pkg
  • bb04d77bda3ae9c9c3b6347f7aef19ac .loader
  • 3efeccfc6daf0bf99dcb36f247364052 4_5983241673595946132.dmg
  • cb56955b70c87767dee81e23503086c3 WbBot.pkg
  • b63e8d4277b190e2e3f5236f07f89eee .loader
  • be37637d8f6c1fbe7f3ffc702afdfe1d MarkMakingBot.dmg
  • bb66ab2db0bad88ac6b829085164cbbb BitcoinTrader.pkg
  • 267a64ed23336b4a3315550c74803611 .loader
  • 6588d262529dc372c400bef8478c2eec UnionCryptoTrader.dmg
  • 55ec67fa6572e65eae822c0b90dc8216 UnionCryptoTrader.pkg
  • da17802bc8d3eca26b7752e93f33034b .unioncryptoupdater
  • 39cdf04be2ed479e0b4489ff37f95bbe JMTTrader_Mac.dmg
  • e35b15b2c8bb9eda8bc4021accf7038d JMTTrader.pkg
  • 6058368894f25b7bc8dd53d3a82d9146 .CrashReporter

Windows malware

  • a9e960948fdac81579d3b752e49aceda WFCUpdater.exe
  • 24B3614D5C5E53E40B42B4E057001770 UnionCryptoTraderSetup.exe
  • 629B9DE3E4B84B4A0AA605A3E9471B31 UnionCryptoUpdater.exe
  • E1953FA319CC11C2F003AD0542BCA822 AdobeUpdator.exe, AdobeARM.exe
  • f221349437f2f6707ecb2a75c3f39145 rasext.dll
  • 055829E7600DBDAE9F381F83F8E4FF36 UnionCryptoTraderSetup.exe
  • F051A18F79736799AC66F4EF7B28594B Unistore.exe

File path

  • %SYSTEM%system32rasext.dll
  • %SYSTEM%system32msctfp.dat
  • %APPDATA%LenovodevicecenterDevice.exe
  • %APPDATA%LenovodevicecenterCenterUpdater.exe
  • %APPDATA%LocalunioncryptotraderUnionCryptoUpdater.exe
  • $APPDATA%adobeAdobeUpdator.exe
  • C:Programdataadobeadobeupdator.exe
  • %AppData%LocalCommsUnistore.exe

Domains and IPs

Domains

  • www.wb-bot.org
  • www.jmttrading.org
  • cyptian.com
  • beastgoc.com
  • www.private-kurier.com
  • www.wb-invest.net
  • wfcwallet.com
  • chainfun365.com
  • www.buckfast-zucht.de
  • invesuccess.com
  • private-kurier.com
  • aeroplans.info
  • mydealoman.com
  • unioncrypto.vip

IPs

  • 104.168.167.16
  • 23.254.217.53
  • 185.243.115.17
  • 104.168.218.42
  • 95.213.232.170
  • 108.174.195.134
  • 185.228.83.32
  • 172.81.135.194

URLs

  • https://www.wb-bot[.]org/certpkg.php
  • http://95.213.232[.]170/ProbActive/index.do
  • http://beastgoc[.]com/grepmonux.php
  • https://unioncrypto[.]vip/update


Temp Mails (https://tempemail.co/) is a new free temporary email addresses service. This service provide you random 10 minutes emails addresses. It is also known by names like: temporary mail, disposable mail, throwaway email, one time mail, anonymous email address… All emails received by Tempmail servers are displayed automatically in your online browser inbox.

Cyberthreats to financial institutions 2020: Overview and predictions – 10 minute mail

Key events 2019

  • Large-scale anti-fraud bypass: Genesis digital fingerprints market uncovered
  • Multi-factor authentication (MFA) and biometric challenges
  • Targeted attack groups specializing in financial institutions: splitting and globalization
  • ATM malware becomes more targeted
  • Card info theft and reuse: magecarting everywhere and battle of POS malware families in Latin America

Large-scale anti-fraud bypass: Genesis digital fingerprints market uncovered

During the last few years, cybercriminals have invested a lot in methods to bypass anti-fraud systems, because now it’s not enough just to steal the login, password and PII – they now need a digital fingerprint to bypass anti-fraud systems in order to extract money from the bank. During 2019, we identified a huge underground market called Genesis, which sells digital fingerprints of online banking users from around the globe.

From an anti-fraud system perspective, the user’s digital identity is a digital fingerprint – a combination of system attributes that are unique to each device, and the personal behavioral attributes of the user. It includes the IP address (external and local), screen information (screen resolution, window size), firmware version, operating system version, browser plugins installed, time zone, device ID, battery information, fonts, etc. The device may have over 100 attributes used for browsing. The second part of a digital identity is the behavioral analysis.

As criminals are continuously looking for ways to defeat anti-fraud safeguards, they try to substitute the system’s real fingerprint with a fake one, or with existing ones stolen from someone else’s PC.

The Genesis Store is an online invitation-only private cybercriminal market for stolen digital fingerprints. At the time of our research, it offered more than 60 thousand stolen bot profiles. The profiles include browser fingerprints, website user logins and passwords, cookies, credit card information, etc. By uploading this fingerprint to the Tenebris Linken Sphere browser, criminals are able to masquerade as legitimate online banking users from any region, country, state, city, etc.

This type of attack shows that criminals have in-depth knowledge of how internal banking systems work and it’s a real challenge to protect against such attacks. The best option is to always use multi-factor authentication.

Multi-factor authentication (MFA) and biometric challenges

MFA is a challenge for cybercriminals. When MFA is used, they have to come up with techniques to bypass it. The most common methods used during the last year were:

  • Exploiting vulnerabilities and flaws in the configuration of the system. For example, criminals were able to find and exploit several flaws in remote banking systems to bypass OTPs (one time passcodes);
  • Using social engineering, a common method among Russian-speaking cybercriminals and in APAC region;
  • SIM swapping, which is especially popular in regions like Latin America and Africa. In fact, despite SMS no longer being considered a secure 2FA, low operational costs mean it’s the most popular method used by providers.

In theory, biometrics should solve a lot of problems associated with two-factor authentication, but practice has shown that it may not be so simple. Over the past year, several cases have been identified that indicate biometrics technology is still far from perfect.

Firstly, there are quite a few implementation problems. For example, Google Pixel 4 does not check if your eyes are open during the unlocking process using facial characteristics. Another example is the possibility of bypassing fingerprint authentication using the sensor under the screen on smartphones made by various manufacturers, including popular brands such as Samsung.

There is another trick that has been exploited in Latin America: a visual capturing attack. Cybercriminals installed rogue CCTV cameras and used them to record the PINs people used to unlock their phones. Such a simple technique is still very effective for both types of victims: those who use biometrics and those who prefer PINs to fingerprints or facial recognition. This is because, when a device is dusty or greasy (and the same applies to a user’s fingers), the best way to unlock a phone is to use a PIN.

Secondly, there were several high-profile leaks of biometric databases. The most notorious was the leak of the Biostar 2 database that included the biometric data of over 1 million people. The company stored unencrypted data, including names, passwords, home addresses, email addresses and, most importantly, unencrypted biometric data that included fingerprints and facial recognition patterns as well as the actual photos of faces. A similar leak occurred at a US Customs and Border Patrol contractor, where biometric information of over 100,000 people was leaked.

There have already been several proof-of-concept attacks that use biometric data to bypass security controls, but those attacks could still be countered with system updates. With these latest leaks, on the other hand, this won’t work because your biometric data cannot be changed – it stays with you forever.

The cases mentioned above, combined with the high-quality research carried out by cybercriminals to obtain a complete digital fingerprint of a user in order to bypass anti-fraud systems, suggest that relying solely on biometric data will not solve the current problems. Today’s implementations need a lot of effort and more research to make them truly secure.

Targeted attack groups specializing in financial institutions: splitting and globalization

FIN7

In 2018, Europol and the US Department of Justice announced the arrest of the leader of the FIN7 and Carbanak/CobaltGoblin cybercrime groups. Some believed that the arrest would have an impact on the group’s operations, but this does not seem to have been the case. In fact, the number of groups operating under the umbrella of CobaltGoblin and FIN7 has grown: there are several interconnected groups using very similar toolkits and the same infrastructure to conduct their cyberattacks.

The first operating under this umbrella is the now-notorious FIN7 that specializes in attacking various companies to get access to financial data or their PoS infrastructure. It relies on the Griffon JScript backdoor and Cobalt/Meterpreter and, in more recent attacks, PowerShell Empire.

The second is CobaltGoblin/Carbanak/EmpireMonkey. It uses the same toolkit, techniques and a similar infrastructure, but targets only financial institutions and associated software and service providers.

The final group is the newly discovered CopyPaste group, which has targeted financial entities and companies in one African country – leading us to believe that this group is associated with cyber-mercenaries or a training center. The links between CopyPaste and FIN7 are still very weak. It’s possible that the operators of this cluster of activity were influenced by open-source publications and don’t actually have any ties to FIN7.

All of these groups benefit greatly from unpatched systems in corporate environments and continue to use effective spear-phishing campaigns in conjunction with well-known Microsoft Office exploits generated by their exploitation framework. So far, the groups have not used any zero-day exploits. FIN7/Cobalt phishing documents may seem basic, but when combined with their extensive social engineering and focused targeting, they have proved to be quite successful.

In the middle of 2019, FIN7 fell silent, but returned at the end of the year with new attacks and new tools. We suspect that the silent period is connected to their infrastructure shutdown that occurred after closing a bulletproof hosting company in Eastern Europe.

In contrast to FIN7, the activity of the Cobalt Goblin Group was stable throughout the year, which once again proves that these groups are connected, but operate on their own: their toolsets and TTPs are very similar, but operate independently; and only occasionally can we spot overlaps in infrastructure. At the same time, the intensity of attacks is slightly lower than in 2018. Cobalt Goblin’s tactics have remained the same: they use documents with exploits that first load a small downloader and then a Cobalt beacon. The main targets also remain the same: small banks in a variety of countries. Perhaps we have detected a lower number of attacks due to diversification, because some indicators suggest the group could also be engaging in JS sniffing (MageCarting) in order to obtain data about payment cards directly from websites.

JS sniffing was extremely popular throughout the year and we found thousands of e-commerce websites infected with these scripts. The injected scripts act in different ways and the infrastructure of the attackers is very different, which suggests that this type of fraud is used by at least a dozen cybercrime groups.

The Silence group actively expanded its operations into different countries throughout the year. We detected attacks in regions where we have never seen them before. For example, we recorded attacks in Southeast Asia and Latin America. This indicates that they have either expanded their operations themselves or started cooperating with other regionally installed cybercrime groups. However, when we look at the development of their main backdoor, we see that their technologies have barely changed over the last two years.

ATM malware becomes more targeted

When it came to ATM malware, we discovered a number of completely new families in 2019. The most notable were ATMJadi and ATMDtrack.

ATMJadi is an interesting one because it doesn’t use the standard XFS, JXFS or CSC libraries. Instead, it uses the victim bank’s ATM software Java proprietary classes: meaning the malware will only work on a small subset of ATMs. It makes this malware very targeted (towards one specific bank).

This is reminiscent of the FASTcach case from 2018, when criminals targeted servers running AIX OS. With a decrease in the number of general-purpose cashout tools, we can say that ATM malware is becoming rarer and more targeted.

Another interesting piece of malware is ATMDtrack, which was first detected in financial institutions in India and is programmed to cash out ATMs. Using the Kaspersky Targeted Attack Attribution Engine (KTAE), we were able to attribute these attacks to the Lazarus group, which supports our prediction from 2018 that there will be “more nation-state sponsored attacks against financial organizations“. Moreover, similar spyware has been found in research centers, with Lazarus APT group using almost identical tools to steal research results from scientific institutes.

Card info theft and reuse

During the year we saw a lot of malware targeting end users and businesses looking for credit card data. In Brazil, in particular, we saw a couple of malware families fighting it out between themselves to maintain control of infected devices. HydraPOS and ShieldPOS were very active during the year, with new versions that included a lot of new targets; Prilex, meanwhile, reduced its activities in the second half of the year.

ShieldPOS has been active since at least 2017 and, after being malware only, it has finally evolved into a MaaS (malware-as-a-service). This fact shows there’s great interest from Latin American cybercriminals in running their own “business” to steal credit cards. HydraPOS has been mostly focused on stealing money from POS systems in restaurants, parking slot machines and different retail stores.

Compared to ShieldPOS, HydraPOS is an older campaign from an actor we named Maggler, which has been in the credit card business since at least 2016. The main difference is that, unlike ShieldPOS, it doesn’t work as MaaS. In both cases, we suspect that the initial infection vector is a carefully prepared social engineering campaign involving telephone calls to the victims.

Analysis of forecasts for 2019

Before giving our forecasts for 2020, let’s see how accurate our forecasts for 2019 turned out to be:

The emergence of new groups due to the fragmentation of Cobalt/Carbanak and FIN7: new groups and new geography.

  • Yes, we saw CobaltGoblin activity, FIN7 activity, CopyPaste activity and the intersection of IoCs and the Silence group.

The first attacks through the theft and use of biometric data.

  • Yes, hacking of various biometric data databases regularly appeared throughout the year. We also revealed a digital fingerprint market where criminals can buy digital fingerprints, which includes, among other things, behavioral data (component of biometrics).

The emergence of new local groups attacking financial institutions in the Indo-Pakistan region, Southeast Asia and Central Europe.

  • No. It turned out that well-known groups such as Lazarus, Silence and CobaltGoblin took their place and very actively attacked financial institutions in these regions.

Continuation of supply-chain attacks: attacks on small companies that provide their services to financial institutions around the world.

Traditional cybercrime will focus on the easiest targets and bypass anti-fraud solutions: replacement of POS attacks with attacks on systems accepting online payments (Magecarting/JS skimming).

  • Yes, the number of groups that started carrying out attacks on online payment systems grew constantly over the year. We detected thousands of websites that were affected by JS skimming.

The cybersecurity systems of financial institutions will be bypassed using physical devices connected to the internal network.

  • Yes, and not only in financial institutions but even the aerospace industry, namely NASA, has suffered from this type of attack.

Attacks on mobile banking for business users.

Advanced social engineering campaigns targeting operators, secretaries and other internal employees in charge of wire transfers.

  • Yes, BEC (business email compromise) attacks have been on the rise worldwide. We have seen major attacks in Japan, while there have also been campaigns in South America, particularly in Ecuador.
  • Additionally, advanced social attacks have been actively used in Brazil to make POS operators go to a malicious website to download specially crafted remote control modules and run them, for example, in HydraPOS attacks.

Forecast 2020

Attacks against Libra and TON/Gram

The successful launch of cryptocurrencies such as Libra and Gram might lead to the worldwide spread of this type of asset, which naturally will attract the attention of criminals. Given the serious surge in cybercriminal activity during the rapid growth of Bitcoin and altcoins in 2018, we predict that a similar situation will most likely unfold around Gram and Libra. Large players in this market should be especially careful, as there are a number of APT groups, such as WildNeutron and Lazarus, whose interests include crypto assets. They are very likely to exploit these developments.

Reselling bank access

During 2019, we witnessed cases where groups who specialize in targeted attacks on financial institutions appeared in the victims’ networks after intrusions by other groups that specialize in selling rdp/vnc access, such as FXMSP and TA505. These facts are also confirmed by underground forums and chat monitoring.

In 2020, we expect an increase in the activity of groups specializing in the sale of network access in the African and Asian regions, as well as in Eastern Europe. Their prime targets are small banks, as well as financial organizations recently bought by big players who are rebuilding their cybersecurity system in accordance with the standards of their parent companies.

Ransomware attacks against banks

This forecast logically follows from the previous one. As mentioned above, small financial institutions often become the victims of opportunistic cybercriminals. If these criminals cannot resell access, or even if it becomes less likely that they will be able to withdraw money, then the most logical monetization of such access is ransomware. Banks are among those organizations that are more likely to pay a ransom than accept the loss of data, so we expect the number of such targeted ransomware attacks to continue to rise in 2020.

Another ransomware attack vector against small and medium financial institutions will be a “pay-per-install” scheme. Traditional botnets will eventually turn into increasingly popular delivery mechanisms against those financial institutions.

2020: the return of custom tooling

Measures taken by antivirus products to effectively detect open source tools used for pen testing purposes, and the adoption of the latest cyberdefense technologies, will push cybercrime actors to return to custom tooling in 2020 and also invest in new Trojans and exploits.

Global expansion of mobile banking Trojans: result of leaked source

Our research and monitoring of underground forums suggests that the source code of some popular mobile banking Trojans was leaked into the public domain. Given the popularity of such Trojans, we expect a repeat of the situation when the source code of ZeuS and SpyEye Trojans were leaked: the number of attempts to attack users will increase at times, and the geography of attacks will expand to almost every country in the world.

Investment apps on the rise: new target for criminals

Mobile investment apps are becoming more popular among users around the globe. This trend won’t go unnoticed by cybercriminals in 2020. Given the popularity of some fintech companies and exchanges (for both real and virtual money), cybercriminals will realize that not all of them are prepared to deal with massive cyberattacks, as some apps still lack basic protection for customer accounts, and do not offer two-factor authentication or certificate pinning to protect app communication. Several governments are deregulating this area and new players are appearing every day, becoming popular very quickly. In fact, we have already seen attempts by cybercriminals to substitute the interfaces of these apps with their own malicious versions.

Magecarting 3.0: even more attacker groups and cloud apps to become prime targets

Over the past couple of years, JS skimming has gained immense popularity among attackers. Unfortunately, cybercriminals now have a huge attack surface that consists of vulnerable e-commerce websites and extremely cheap JS skimmer tools available for sale on various forums, starting at $200. At the moment we are able to distinguish at least 10 different actors involved in these types of attacks and we believe that their number will continue to grow during the next year. The most dangerous attacks will be on companies that provide services such as e-commerce as a service, which will lead to the compromise of thousands of companies.

Political instability leading to the spread of cybercrime in specific regions

Some countries are experiencing political and social upheaval, resulting in masses of people seeking refugee status in other countries. These waves of immigration include all sorts of people, including cybercriminals. This phenomenon will result in the spread of geographically localized attacks in countries that have not previously been affected by them.


Temp Mails (https://tempemail.co/) is a new free temporary email addresses service. This service provide you random 10 minutes emails addresses. It is also known by names like: temporary mail, disposable mail, throwaway email, one time mail, anonymous email address… All emails received by Tempmail servers are displayed automatically in your online browser inbox.