NOVEMBER 11, 2025
This report provides an in-depth analysis of the 2025 operational patterns and Tactics, Techniques, and Procedures (TTPs) of the Advanced Persistent Threat (APT) group assessed to be Iranian state-sponsored, known in the cybersecurity community as APT33, Elfin, and by Microsoft as Peach Sandstorm. APT33 acts as a dynamic instrument of Iranian state policy, not merely a cybercrime group, and has undergone a strategic evolution in recent years. The central thesis of this report is that APT33’s 2025 threat posture is defined by a “Cloud-First” attack model, representing a paradigm shift from traditional network-based attacks to identity-centric infiltrations.
This evolution constitutes an urgent call to action at the leadership level. Defense investments and focus must shift from the network perimeter to the identity and cloud security layers. Key recommendations include mandating phishing-resistant Multi-Factor Authentication (MFA), hardening cloud configurations, and integrating geopolitical intelligence into cyber risk assessment processes.
There is strong and consistent evidence that APT33 conducts its operations on behalf of the Iranian state, specifically in connection with the Islamic Revolutionary Guard Corps (IRGC). This attribution is based on various technical and operational indicators. The group’s operational hours align with Iran’s Saturday-to-Wednesday work week. Furthermore, traces of the Farsi language have been found in malware code, and Iranian-origin hosting services have been used for infrastructure.
One of the most concrete pieces of evidence is the developer handle “xman_1365_x” found in the PDB (Program Database) path of the group’s custom-developed TURNEDUP backdoor. This handle has been linked to the “Nasr Institute,” which is considered Iran’s “cyber army” and an extension of the IRGC. This connection draws a direct line from a piece of custom malware to the state apparatus, indicating that APT33 is not an ordinary cybercrime group but an organized unit serving Iran’s strategic objectives.
APT33 is tracked by numerous security organizations under different names. This can create challenges in correlating intelligence from different sources. The following table is prepared to clarify these naming conventions and allow a SOC analyst to associate a “Peach Sandstorm” alert from Microsoft Defender with a “Refined Kitten” report from CrowdStrike.
| Alias | Tracking Organization |
| APT33 | Mandiant (Google) |
| Elfin | Symantec (Broadcom) |
| Peach Sandstorm | Microsoft |
| HOLMIUM | Microsoft (former name) |
| Refined Kitten | CrowdStrike |
| MAGNALLIUM | Dragos |
| COBALT TRINITY | SecureWorks |
| G0064 | MITRE ATT&CK |
| TA451 | Proofpoint |
| ATK 35 | Thales |
Table 1: APT33 Alias Correlation Matrix
APT33’s activities are shaped around two main strategic objectives: industrial espionage and destructive sabotage. This dual nature reveals that the group is not a monolithic entity serving a single purpose, but a strategic asset that can be flexibly used according to Iran’s geopolitical needs.
The structure of this group reflects a doctrine of “Operational Duality,” which can be described not as espionage or sabotage, but as espionage for sabotage. The intelligence-gathering phase is critical for mapping networks, identifying high-value targets (like Industrial Control Systems), and pre-positioning access for a future destructive attack. The DROPSHOT dropper, capable of delivering both the TURNEDUP espionage backdoor and the SHAPESHIFT wiper, is tactical proof of this doctrine. This means that detecting an APT33 espionage tool on a network should not be treated merely as a data breach, but as the first stage of a potential sabotage operation. This reframes the incident response priority from data containment to preventing physical disruption.

APT33‘s early operations focused on developing basic infiltration and espionage capabilities. During this period, the main initial access vector was spearphishing emails prepared with recruitment-themed lures. These emails directed victims to click on malicious HTML Application (.hta) files and used domain names impersonating legitimate companies like Boeing to increase credibility. In later years, the group diversified its TTPs, beginning to systematically exploit known vulnerabilities such as CVE-2017-11774 in Microsoft Outlook and CVE-2018-20250 in WinRAR.
This is the most critical evolution, forming the basis of the group’s projected activities for 2025.
This operational model reveals a sophisticated “Cloud Supply Chain” attack model. APT33 has established a supply chain for its C2 infrastructure. They attack “soft targets” like universities with low-cost, high-volume password spray attacks to compromise accounts. They then use these compromised education sector accounts to create new, fraudulent Azure subscriptions (e.g., “Azure for Students”). This “clean” infrastructure, hosted on Microsoft’s trusted platform, is then used to launch stealthy attacks against their actual high-value targets in the defense and energy sectors. This shows that an attack on a university is not the final goal but a logistical step for a larger operation. This proves that organizations in non-target sectors are also at risk of being abused as infrastructure.
APT33’s target selection directly aligns with Iran’s national interests: aerospace, defense, energy (oil, gas, and petrochemicals), and government. This targeting is directly linked to Iran’s geopolitical objectives. For example, targeting Saudi petrochemical companies serves the purpose of gaining a competitive advantage, while attacks on aerospace companies reflect a desire to understand Saudi military capabilities.
The most alarming trend for 2025 is the group’s clear shift in focus to Operational Technology (OT) and Industrial Control Systems (ICS) environments. This forms the basis of the “Cloud-to-OT” attack scenario, indicating the group is no longer just stealing data but has the potential to sabotage physical processes.
APT33 dedicates significant resources to developing its own tools for the most critical phases of its operations.
APT33 heavily leverages existing tools to maximize operational efficiency.
Comprehensive Toolset Inventory Table
| Tool Name | Category | Origin | Key Operational Notes / Associated TTPs |
| Tickler | Multi-stage Backdoor | Custom | Next-gen backdoor discovered in 2024. Uses DLL sideloading and PEB traversal. Targets Azure infrastructure for C2. |
| SHAPESHIFT | Disk Wiper | Custom | Capable of destroying disks and files. Shows strong similarities to Shamoon. |
| DROPSHOT | Dropper | Custom | A tool that drops other malware, such as TURNEDUP or SHAPESHIFT, onto the target system. |
| TURNEDUP | Backdoor | Custom | Capabilities for file upload/download, system info collection, and reverse shell. Associated with “xman_1365_x”. |
| POWERTON | PowerShell Implant | Custom | PowerShell-based implant with encrypted C2, multiple persistence mechanisms, and ability to dump password hashes. |
| Remcos RAT | Remote Access Tool | Commercial | A commonly used Remote Access Trojan (RAT). Preferred for its broad functionality. |
| DarkComet RAT | Remote Access Tool | Commercial | Placed in the Startup folder for persistence. |
| Quasar RAT | Remote Access Tool | Open Source | An open-source and widely used RAT. |
| Mimikatz | Credential Dumping | Open Source | Used to extract plaintext passwords and hashes from memory (LSASS) on Windows systems. |
| LaZagne | Credential Dumping | Open Source | Collects passwords stored in browsers, email clients, and other applications. |
| PowerShell Empire | C2 Framework | Open Source | A PowerShell-based framework used for post-exploitation tasks, lateral movement, and C2 communication. |
Table 2: APT33 Tool and Malware Arsenal
APT33‘s C2 infrastructure adopts an infrastructure-level camouflage strategy to avoid detection. Defense models based on blocking traditionally known malicious IP addresses or domains are rendered ineffective by the group’s hosting of its C2 infrastructure on legitimate, high-reputation cloud platforms. This approach is a fundamental attack on the trust model of corporate IT, as organizations cannot block entire cloud services like azurewebsites.net without disrupting critical business functions. This forces defenders to move away from simple blocklists and toward more complex, behavioral analytics capable of detecting anomalous behavior within trusted channels. Observed C2 domains include examples like subreviews.azurewebsites[.]net and satellite2.azurewebsites[.]net.
The group typically uses HTTP/S for C2 communication, favoring non-standard ports like TCP 808 and 880 to bypass simple firewall rules. The content of the C2 traffic is encrypted with AES and Base64 encoded to hide it from network-based inspection systems.
APT33’s data exfiltration strategy demonstrates a conscious compartmentalization of operational risk. The group often uses a separate protocol, such as unencrypted FTP (T1048.003), for large data transfers. This may seem illogical for a sophisticated actor at first glance. However, it is a deliberate tactical choice. The C2 channel is the most valuable asset for maintaining persistence and must remain “low and slow.” Exfiltrating gigabytes of data through this channel would create a large traffic anomaly, which could be easily detected. By using a separate and expendable FTP server, they isolate the risk of the “noisy” data transfer from the C2 channel. If the FTP traffic is detected, the attackers only lose the stolen data and the FTP server, but their primary, stealthy C2 channels remain intact. This is a calculated risk where data security (encryption) is sacrificed for operational security (stealth).
The following matrix maps APT33’s observed TTPs to the industry-standard framework, providing an actionable basis for defenders to understand the group’s behavioral patterns and develop specific detection strategies against them.
| Tactic | Technique ID and Name | APT33 Implementation Details |
| Initial Access | T1110.003: Brute Force: Password Spraying | Conducts slow and low password spray attacks against Microsoft 365 and Azure AD tenants. Known to use the ‘go-http-client’ user agent in attacks. |
| T1566: Phishing | Uses spearphishing emails. These contain either archive attachments like .rar (CVE-2018-20250) or links directing to .hta files. | |
| T1078.004: Valid Accounts: Cloud Accounts | Gains initial access using cloud accounts compromised through password spray attacks. | |
| Execution | T1059.001: Command and Scripting Interpreter: PowerShell | Heavily uses PowerShell to download files from the C2 server, run various scripts, and initiate reverse shell sessions. |
| Persistence | T1547.001: Boot or Logon Autostart Execution: Registry Run Keys / Startup Folder | Places RATs like DarkComet in the Startup folder. Uses Registry Run keys for persistence, including for new software like Tickler. |
| T1053.005: Scheduled Task/Job: Scheduled Task | Creates scheduled tasks to run a malicious .vbe file multiple times a day. | |
| Credential Access | T1003: OS Credential Dumping | Uses various tools like LaZagne, Mimikatz, and ProcDump to dump passwords and hashes from LSASS memory, LSA Secrets, and cached domain credentials. |
| Command and Control | T1071.001: Application Layer Protocol: Web Protocols | Uses HTTP for C2 communication. Often prefers non-standard ports like 808 and 880. |
| Impact | T1485: Data Destruction | Forms the basis of the group’s destructive capability. The SHAPESHIFT wiper and links to Shamoon attacks are proof of this tactic. |
Table 3: Mapping APT33 TTPs to MITRE ATT&CK®
APT33‘s TTPs demonstrate an adoption of a philosophy that can be called “Living off the Land and Clouds.” This approach relies on the attacker abusing legitimate tools and services already present in the target environment to conceal their activities.
This section provides the most actionable part of the report, giving security teams the specific data needed for detection and blocking. The following indicators are the most current and actionable collection compiled from the provided research materials.
| Indicator | Type | Associated Malware/TTP | Notes |
| subreviews.azurewebsites[.]net | Domain Name | Tickler | Hosted on a fraudulent Azure subscription |
| satellite2.azurewebsites[.]net | Domain Name | Tickler | Hosted on a fraudulent Azure subscription |
| nodetestservers.azurewebsites[.]net | Domain Name | Tickler | Hosted on a fraudulent Azure subscription |
| satellitegardens.azurewebsites[.]net | Domain Name | Tickler | Hosted on a fraudulent Azure subscription |
| softwareservicesupport.azurewebsites[.]net | Domain Name | Tickler | Hosted on a fraudulent Azure subscription |
| boeing.servehttp[.]com | Domain Name | Spearphishing | Domain Masking |
| “go-http-client” | User Agent | Password Spraying (T1110.003) | Should be monitored in authentication logs |
Table 4: Network-Based Indicators of Compromise
| Indicator | Type | Associated Malware/TTP | Notes |
| 7eb2e9e8cd450fc353323fd2e8b84fbbdfe061a8441fd71750250752c577d198 | SHA-256 | Tickler | YAHSAT NETWORK_…GUIDE_20240421.pdf.exe |
| ccb617cc7418a3b22179e00d21db26754666979b4c4f34c7fda8c0082d08cec4 | SHA-256 | Tickler | Sold.dll |
| 5df4269998ed79fbc997766303759768ce89ff1412550b35ff32e85db3c1f57b | SHA-256 | Tickler | .batch file (persistence) |
| fb70ff49411ce04951895977acfc06fa468e4aa504676dedeb40ba5cea76f37f | SHA-256 | Tickler | .dll file (backdoor) |
| HKCU\Software\Microsoft\Office<Version>\Outlook\WebView\Inbox “URL” = http://… | Registry Key | Ruler / Outlook Persistence | Persistence targeting the Exchange client |
| %LOCALAPPDATA%\SmartMega.exe | File Path | APT33 Backdoor | Known malware location |
| %APPDATA%\MsdUpdate.exe | File Path | APT33 Backdoor | Known malware location |
Table 5: Host-Based Indicators of Compromise
APT33‘s most dangerous attack path for 2025 is a hybrid scenario involving the following steps: 1) Compromising an engineer’s cloud-based Microsoft 365 account via password spraying. 2) Using this identity to infiltrate the corporate IT network. 3) Moving laterally from the IT network to the sensitive OT/ICS network. This scenario represents a paradigm shift in the threat model for critical infrastructure, where a simple password weakness in the cloud can lead to the disruption of physical industrial processes. The following recommendations are designed to prevent this scenario.

Take control of your digital security with an exclusive demo of our powerful threat management platform.