Recent attacks against point-of-sale (PoS) thin clients around the world have been using card data scraping malware and the Cobalt Strike beacon, security firm Morphisec reveals.
The attacks bear marks of the FIN6 hacking group, albeit with some differences, yet could also be tied to the EmpireMonkey group, and the security researchers say they don’t have enough data for proper attribution yet.
The attackers used the FrameworkPOS scraping malware to target some systems, but also used PowerShell/WMI stages to download and load Cobalt Strike with PowerShell extensions directly into memory. Some of the attacks seemed to expressly target PoS VMware Horizon thin clients.
The Cobalt Strike beacon payload gives attackers control over the infected system and also allows them to move laterally to other systems on the network. It can also be used to harvest user credentials, execute code, and perform other malicious operations.
Victims of these attacks were identified across the United States, Japan, and India, from the finance, insurance and healthcare (diagnostic image processing) sectors. Other targets were observed as well.
The researchers believe that one of the infection vectors in these attacks involved HTA (HTML Application) files that execute PowerShell scripts as part of an embedded VBScript. Other scripts leading to the same Cobalt Strike beacon were also identified, but it’s unclear whether they were used for infiltration or lateral movement.
Two types of beacons were observed during the campaign. One is a regular direct reflective loaded Cobalt Strike DLL beacon, while the other is a shellcode backdoor beacon with PowerShell and Mimikatz functionality.
In some attacks, after the backdoors are executed, a task is created to execute and export a function from a downloaded binary DLL with system privileges. The DLL is a FrameworkPOS scraper sample similar to a previously seen FrameworkPOS used by FIN6, designed to exfiltrate data through DNS tunneling.
“These types of advanced attacks that utilize memory to evade detection solutions either by reflectively loading libraries, hollowing process memory or injecting code into new processes, are harder and harder to attribute due to the simple fact that more and more criminals are taking advantage of the strength of these evasion techniques and the weakness of runtime detection technologies to cope with such evasion,” Morphisec concludes.