Targeted Attack on Government Entities in the Middle East | Part 1

Introduction

In July 2026, Zscaler ThreatLabz observed new activity by a threat actor with links to East Asia targeting government entities in the Middle East. During analysis, ThreatLabz captured post-compromise activity and uncovered previously undocumented malware tooling, including TELESHIM, MIXEDKEY, and BINDCLOAK. The campaign used a multi-stage attack chain to establish and maintain access on infected systems, with TELESHIM abusing the Telegram API for command-and-control (C2) communication to blend in with legitimate internet traffic.

This blog post (Part 1) explores the technical details of the multi-stage attack chain, focusing on TELESHIM, MIXEDKEY, and the post-compromise activity observed during the campaign. A follow-up post (Part 2) will provide a detailed technical analysis of the BINDCLOAK C2 implant.

Key Takeaways

• In July 2026, ThreatLabz observed activity by a threat actor linked to East Asia targeting government entities in the Middle East.
• ThreatLabz identified full post-compromise activity leading to the discovery of previously undocumented malware tooling.
• The campaign used a multi-stage attack chain to deploy TELESHIM, MIXEDKEY, and BINDCLOAK on infected systems.
• TELESHIM and MIXEDKEY used heavy code obfuscation techniques leveraging control flow flattening (CFF), mixed boolean arithmetic (MBA), and opaque predicates to hinder reverse engineering. TELESHIM also used various techniques to detect the presence of virtualization-based analysis environments.
• The threat actor leveraged environmental keying by encrypting BINDCLOAK using a decryption key derived from the infected machine’s volume serial number. This helps ensure the payload decrypts and executes only on intended targets.
• TELESHIM abused the Telegram API for C2 communication to blend in with legitimate internet traffic.

Technical Analysis

In the following sections, ThreatLabz provides a technical analysis of the campaign, detailing the malware tooling it leverages and the post-compromise activity.

The attack chain begins with an ISO file containing a legitimate RegSchdTask.exe file  from ASUSTek that sideloads a malicious DLL named AsTaskSched.dll. The figure below illustrates the campaign’s attack flow used to distribute TELESHIM, MIXEDKEY, and BINDCLOAK.

Figure 1: Multi-stage attack chain leading to the deployment of BINDCLOAK.

TELESHIM backdoor

TELESHIM is a 32-bit C++ Windows DLL that is used in the first stage of the attack. ThreatLabz identified three unique instances of TELESHIM. While two of these instances were compiled in 2025, the variant used in this campaign was compiled in July 2026. This variant introduced heavy code obfuscation techniques as well as encrypted strings.

Hook installer (indirect execution)

When the DllMain is invoked, the code locates the base address of the host executable and installs the following 7-byte hook at offset 0x1394 as shown below.

Figure 2: 7-byte hook at offset 0x1394.

Before installing the hook, DllMain calls VirtualProtect to mark this memory region as writable and restores the original protection after writing the hook. The offset 0x1394 is hardcoded, indicating that the attacker reverse-engineered the legitimate host application to identify an instruction executed early on during its normal initialization. When the host executable reaches that point in its startup, the trampoline fires and execution enters the implant's main payload. By redirecting execution via an installed hook, the code is executed in the context of the host executable to evade security tools.

Mutex check

TELESHIM checks for the presence of a mutex named ----WebKitFormBoundary7MA4YWxkTrZu0g on the victim’s machine to ensure only a single instance of the malware is running at any given time.

String decryption

All strings relevant to the malicious activity are stored encrypted in the binary. Below are two string decryption variants.

Method 1

In this variant, each string has its own dedicated CFF-obfuscated decryption function. The XOR key and ciphertext are stored contiguously in a single .rdata blob. Every function is structurally identical; only the decryption key length and plaintext length operands change.

decryption_key = blob[:key_len]
ciphertext  = blob[key_len:]

plaintext = bytearray(plaintext_len)
for i in range(plaintext_len):
   plaintext[i] = decryption_key[i % key_len] ^ ciphertext[i]

Method 2

In this variant, each string is first Base64-decoded and then decrypted using the following 44-byte rolling XOR key.

8F 38 0C DA 29 6F 34 DE 27 69 7A 1A 53 05 18 49 B6 9D 59 E5 28 D7 E6 
69 F1 7C F8 D3 CF 22 0B 66 96 DA 77 65 34 40 1C 8A 0F 0C 31 C6

This string decryption algorithm is also used to protect the network communication, which is discussed in more detail later.

Anti-analysis techniques

TELESHIM uses multiple anti-analysis techniques to evade automated analysis environments and hinder static reverse engineering.

I/O file stress test

TELESHIM delays execution by performing intense I/O file activity using a function that writes and reads back ~1 MB of randomly generated data to %TEMP%\CVR9EEA.tmp in a loop of 1000 iterations, generating roughly 1 GB of total disk I/O per invocation. This is likely to stall the execution in emulated, virtual, and analysis environments.

Hypervisor detection using CPUID

The next anti-analysis check calls the CPUID instruction with EAX=1 and checks bit 31 of the ECX register to detect the presence of a hypervisor. Since all major hypervisors set this bit, it can be used to detect virtualization-based analysis environments. If bit 31 of ECX is set,  execution terminates.

RAM speed check using WMI

TELESHIM leverages Windows Management Instrumentation (WMI) to execute the following WMI query to check RAM speed.

wmic memorychip get speed

In a virtualization environment, this command usually returns either 0 or an undefined value. If the return value is 0 or cannot be parsed, then the execution terminates. The WMI query string itself is stored encrypted using string decryption method 2.

CFF and MBA usage

TELESHIM uses CFF and complex MBA expressions to deter reverse engineering. The figure below shows the CPUID-based anti-analysis technique leveraging MBA expressions.

Figure 3: MBA expressions used to obfuscate the CPUID hypervisor bit check in TELESHIM.

Similar MBA expressions are used in other critical sections of the code, such as string decryption.

Staging payloads

TELESHIM creates a directory at C:\programdata\shimgen_Data\ for staging payloads. The legitimate executable (RegSchdTask.exe) is copied to this path as shimgen.exe and the malicious DLL is copied to this path as AsTaskSched.dll.

Persistence via scheduled tasks

TELESHIM creates a scheduled task named shimgen that runs every 6 minutes and executes the binary from C:\programdata\shimgen_Data\shimgen.exe.

C2 communication

TELESHIM abuses the Telegram API for C2 communication, a technique used to blend in with legitimate internet traffic. The hostname api.telegram.org, the Telegram bot token, and the chat ID are stored encrypted and encoded in the binary using string encryption method 2. These values are decrypted at runtime to enable C2 communication.

The GetAdaptersInfo API is used to retrieve the infected machine's MAC address, which uniquely identifies the machine to the C2 server.

In the next phase, TELESHIM enters a polling loop by sending HTTP GET requests to the following URL to fetch updates: https://api.telegram.org/bot[BOT_TOKEN]/getUpdates?offset=[N].

Each HTTP GET request uses the hardcoded User-Agent below to impersonate web browser traffic: 

Mozilla/5.0 (Macintosh; Intel Mac OS X 10_5_8) AppleWebKit/534.31 (KHTML, like Gecko) Chrome/13.0.748.0 Safari/534.31

The JSON response from the Telegram server is parsed using the nlohmann::json library written in C++ to extract the relevant fields.

Chat ID validation

The message.chat.id field parsed from the JSON response is validated against the hardcoded chat_id in the binary to prevent hijacking.

C2 message types

There are two C2 message types supported by TELESHIM that are processed with different handlers, which include the following:

• Control messages:
  • TELESHIM decrypts message.text using string decryption method 2. If the decrypted message is the string 13, TELESHIM XOR-encrypts and Base64-encodes the machine’s MAC address and sends it in response to the Telegram bot to register the infection.
  • If the decrypted message.text is not the string 13, it is parsed as a command and executed using cmd.exe /C by calling CreateProcessA. TELESHIM only executes commands addressed to its MAC address in the format [MAC_ADDRESS] [SHELL_COMMAND]. The command output is encrypted using method 2 and sent in chunks if it is larger than 1,000 bytes.
• Download and execute messages:
  • TELESHIM decrypts message.caption using method 2; the result is in the format [MAC address] [destination file path]. It validates the MAC address against the machine’s own MAC address and, if it matches, uses document.file_id parsed from the JSON response to fetch the file via /bot[TOKEN]/getFile?file_id=[FILE_ID]. The downloaded file is decrypted using the 44-byte rolling XOR key, and scheduled tasks are used to launch the dropped binary.

Post-compromise activity

At the time of analysis, ThreatLabz captured post-compromise activity from the C2 operator, including system, user, and network reconnaissance commands, along with the deployment of next-stage payloads. Most of the activity took place between July 7, 2026 and July 9, 2026. Using the original timestamps at which the threat actor issued the C2 commands, ThreatLabz performed a timing analysis and observed that all C2 commands were executed only between 4 AM UTC and 12 PM UTC, with a heavy concentration between 7 AM and 11 AM UTC. The figure below shows the time-series plot.

Figure 4: Time plot of C2 commands executed by the TELESHIM threat actor.

The table below summarizes some of the C2 commands executed by the threat actor.

Table 1: Post-compromise commands executed by the TELESHIM threat actor.

Deployment of next stage payloads

For each infected machine, the threat actor enumerated the directories under the path C:\ProgramData\ to choose an appropriate staging directory. The following next-stage payloads were then deployed to the staging directory.

• Legitimate GoPro binary GoProAlertService.exe. For each infected machine, the legitimate binary was dropped with an appropriate name to blend in with the name of the staging directory.
• Legitimate MSVCP120.dll
• Legitimate MSVCR120.dll
• Malicious DLL - pthreadVC2.dll sideloaded by the legitimate binary

On each infected machine, the threat actor created the directory C:\ProgramData\Crypto\DSS\ and dropped the final encrypted C2 implant named C99F29AC08454855B3D538960BB2F34F.PCPKEY. The directory name and the file extension were carefully chosen by the threat actor to impersonate files related to Microsoft’s Platform Crypto Provider (PCP) in order to appear benign.

Once all the next-stage payloads were dropped, the threat actor created a scheduled task named Feedback that runs every 10 minutes and launches the legitimate binary, which sideloads the malicious DLL named pthreadVC2.dll present alongside in the same directory. ThreatLabz named this next stage loader MIXEDKEY.

MIXEDKEY reflective loader

MIXEDKEY is a Windows 64-bit DLL whose main purpose is to decrypt the contents of C99F29AC08454855B3D538960BB2F34F.PCPKEY, and reflectively load it.

Similar to the TELESHIM backdoor, MIXEDKEY heavily uses MBA operations to generate junk instructions and opaque predicates, bloating the size of the binary and deterring reverse engineering.

String decryption

Unlike TELESHIM, MIXEDKEY constructs decrypted strings at runtime by computing each byte using MBA expressions over hardcoded values in the .data section. These bytes are written out of order to a buffer to assemble the final string. To compute a single byte, MIXEDKEY executes approximately ~1,000 instructions.

Final payload decryption

The final payload is encrypted using two layers of XOR encryption. The second layer of encryption uses environmental keying. The threat actor used the volume serial number of the victim's machine to derive a key to encrypt the payload before deploying it. This ensures the payload decrypts and executes only on the intended target and complicates decryption by analysts who do not have the correct volume serial number.

MIXEDKEY decrypts the payload and reflectively loads it using the following steps:

• Fetches the 4-byte volume serial number by calling GetVolumeInformationA.
• Derives a 20-byte rolling XOR key from the 4-byte volume serial number by repeating the 4-byte serial number five times.
• Reads the contents of C:\ProgramData\Crypto\DSS\C99F29AC08454855B3D538960BB2F34F.PCPKEY to load the encrypted payload.
• Uses the first 311 bytes of this file as a rolling XOR key to decrypt the rest of the file.
• The decrypted output of the previous step is once again decrypted using the 20-byte rolling XOR key derived from the volume serial number.
• The final decrypted output contains a portable executable (PE) file with the "MZ" signature stripped off and the 4-byte payload size prefixed to the payload.

Finally, MIXEDKEY reflectively loads the PE file and invokes its export function to continue the next stage of the attack chain.

BINDCLOAK C2 implant

The final payload is a 64-bit C2 implant written in C++ that ThreatLabz tracks as BINDCLOAK. It beacons to the C2 server at cert.hypersnet[.]com. A detailed technical analysis of BINDCLOAK will be shared in a follow-up blog post.

Threat Attribution

During our analysis of the post-compromise activity, ThreatLabz observed high-confidence indicators revealing the threat actor’s public IP address and the system locale configured on their Windows server. Based on the geolocation of the IP address, the configured system locale, and active operational hours matching regional working timeframes, ThreatLabz assesses with moderate-to-high confidence that the threat actor is operating out of East Asia. At this stage, ThreatLabz is not attributing this activity to any known APT group. We will update our attribution as more evidence emerges.

To Be Continued

In this campaign, a threat actor targeted government entities in the Middle East using a multi-stage attack chain with previously undocumented malware tooling including, TELESHIM, MIXEDKEY, and BINDCLOAK. The activity also reflects broader trends such as EDR evasion, blending in with legitimate internet traffic through abuse of trusted platforms, and the use of code-obfuscation techniques such as MBA and CFF to hinder reverse engineering.

This post (Part 1) focused on the attack chain, TELESHIM, MIXEDKEY, and post-compromise activity; a follow-up post (Part 2) will provide a detailed technical analysis of the BINDCLOAK C2 implant.

Zscaler Coverage

Zscaler’s multilayered cloud security platform detects indicators related to this campaign at various levels.

• Win32.Backdoor.TELESHIM
• Win64.Loader.MIXEDKEY

Indicators Of Compromise (IOCs)

File indicators
 

Network indicators