Enterprise sub-techniques

Adversaries may use startup items automatically executed at boot initialization to establish persistence. Startup items execute during the final phase of the boot process and contain shell scripts or other executable files along with configuration information used by the system to determine the execution order for all startup items.[1]

This is technically a deprecated technology (superseded by Launch Daemon), and thus the appropriate folder, /Library/StartupItems isn’t guaranteed to exist on the system by default, but does appear to exist by default on macOS Sierra. A startup item is a directory whose executable and configuration property list (plist), StartupParameters.plist, reside in the top-level directory.

An adversary can create the appropriate folders/files in the StartupItems directory to register their own persistence mechanism.[2] Additionally, since StartupItems run during the bootup phase of macOS, they will run as the elevated root user.

Adversaries may use steganography techniques in order to prevent the detection of hidden information. Steganographic techniques can be used to hide data in digital media such as images, audio tracks, video clips, or text files.

Duqu was an early example of malware that used steganography. It encrypted the gathered information from a victim's system and hid it within an image before exfiltrating the image to a C2 server.[1]

By the end of 2017, a threat group used Invoke-PSImage to hide PowerShell commands in an image file (.png) and execute the code on a victim's system. In this particular case the PowerShell code downloaded another obfuscated script to gather intelligence from the victim's machine and communicate it back to the adversary.[2]

Adversaries may use steganographic techniques to hide command and control traffic to make detection efforts more difficult. Steganographic techniques can be used to hide data in digital messages that are transferred between systems. This hidden information can be used for command and control of compromised systems. In some cases, the passing of files embedded using steganography, such as image or document files, can be used for command and control.

Adversaries may insert, delete, or manipulate data at rest in order to influence external outcomes or hide activity, thus threatening the integrity of the data.[1][2] By manipulating stored data, adversaries may attempt to affect a business process, organizational understanding, and decision making.

Stored data could include a variety of file formats, such as Office files, databases, stored emails, and custom file formats. The type of modification and the impact it will have depends on the type of data as well as the goals and objectives of the adversary. For complex systems, an adversary would likely need special expertise and possibly access to specialized software related to the system that would typically be gained through a prolonged information gathering campaign in order to have the desired impact.

Adversaries may attempt to make a payload difficult to analyze by removing symbols, strings, and other human readable information. Scripts and executables may contain variables names and other strings that help developers document code functionality. Symbols are often created by an operating system’s `linker` when executable payloads are compiled. Reverse engineers use these symbols and strings to analyze code and to identify functionality in payloads.[1][2]

Adversaries may use stripped payloads in order to make malware analysis more difficult. For example, compilers and other tools may provide features to remove or obfuscate strings and symbols. Adversaries have also used stripped payload formats, such as run-only AppleScripts, a compiled and stripped version of AppleScript, to evade detection and analysis. The lack of human-readable information may directly hinder detection and analysis of payloads.[3]

Adversaries may perform sudo caching and/or use the sudoers file to elevate privileges. Adversaries may do this to execute commands as other users or spawn processes with higher privileges.

Within Linux and MacOS systems, sudo (sometimes referred to as "superuser do") allows users to perform commands from terminals with elevated privileges and to control who can perform these commands on the system. The sudo command "allows a system administrator to delegate authority to give certain users (or groups of users) the ability to run some (or all) commands as root or another user while providing an audit trail of the commands and their arguments."[1] Since sudo was made for the system administrator, it has some useful configuration features such as a timestamp_timeout, which is the amount of time in minutes between instances of sudo before it will re-prompt for a password. This is because sudo has the ability to cache credentials for a period of time. Sudo creates (or touches) a file at /var/db/sudo with a timestamp of when sudo was last run to determine this timeout. Additionally, there is a tty_tickets variable that treats each new tty (terminal session) in isolation. This means that, for example, the sudo timeout of one tty will not affect another tty (you will have to type the password again).

The sudoers file, /etc/sudoers, describes which users can run which commands and from which terminals. This also describes which commands users can run as other users or groups. This provides the principle of least privilege such that users are running in their lowest possible permissions for most of the time and only elevate to other users or permissions as needed, typically by prompting for a password. However, the sudoers file can also specify when to not prompt users for passwords with a line like user1 ALL=(ALL) NOPASSWD: ALL.[2] Elevated privileges are required to edit this file though.

Adversaries can also abuse poor configurations of these mechanisms to escalate privileges without needing the user's password. For example, /var/db/sudo's timestamp can be monitored to see if it falls within the timestamp_timeout range. If it does, then malware can execute sudo commands without needing to supply the user's password. Additional, if tty_tickets is disabled, adversaries can do this from any tty for that user.

In the wild, malware has disabled tty_tickets to potentially make scripting easier by issuing echo \'Defaults !tty_tickets\' >> /etc/sudoers.[3] In order for this change to be reflected, the malware also issued killall Terminal. As of macOS Sierra, the sudoers file has tty_tickets enabled by default.

Adversaries may employ a known symmetric encryption algorithm to conceal command and control traffic rather than relying on any inherent protections provided by a communication protocol. Symmetric encryption algorithms use the same key for plaintext encryption and ciphertext decryption. Common symmetric encryption algorithms include AES, DES, 3DES, Blowfish, and RC4.

Adversaries may abuse SyncAppvPublishingServer.vbs to proxy execution of malicious PowerShell commands. SyncAppvPublishingServer.vbs is a Visual Basic script associated with how Windows virtualizes applications (Microsoft Application Virtualization, or App-V).[1] For example, Windows may render Win32 applications to users as virtual applications, allowing users to launch and interact with them as if they were installed locally.[2][3] The SyncAppvPublishingServer.vbs script is legitimate, may be signed by Microsoft, and is commonly executed from `\System32` through the command line via `wscript.exe`.[4][5]

Adversaries may abuse SyncAppvPublishingServer.vbs to bypass PowerShell execution restrictions and evade defensive counter measures by "living off the land."[6][4] Proxying execution may function as a trusted/signed alternative to directly invoking `powershell.exe`.[7]

For example, PowerShell commands may be invoked using:[5]

`SyncAppvPublishingServer.vbs "n; {PowerShell}"`

Adversaries may employ various system checks to detect and avoid virtualization and analysis environments. This may include changing behaviors based on the results of checks for the presence of artifacts indicative of a virtual machine environment (VME) or sandbox. If the adversary detects a VME, they may alter their malware to disengage from the victim or conceal the core functions of the implant. They may also search for VME artifacts before dropping secondary or additional payloads. Adversaries may use the information learned from Virtualization/Sandbox Evasion during automated discovery to shape follow-on behaviors.[1]

Specific checks will vary based on the target and/or adversary, but may involve behaviors such as Windows Management Instrumentation, PowerShell, System Information Discovery, and Query Registry to obtain system information and search for VME artifacts. Adversaries may search for VME artifacts in memory, processes, file system, hardware, and/or the Registry. Adversaries may use scripting to automate these checks into one script and then have the program exit if it determines the system to be a virtual environment.

Checks could include generic system properties such as host/domain name and samples of network traffic. Adversaries may also check the network adapters addresses, CPU core count, and available memory/drive size. Once executed, malware may also use File and Directory Discovery to check if it was saved in a folder or file with unexpected or even analysis-related naming artifacts such as `malware`, `sample`, or `hash`.

Other common checks may enumerate services running that are unique to these applications, installed programs on the system, manufacturer/product fields for strings relating to virtual machine applications, and VME-specific hardware/processor instructions.[2] In applications like VMWare, adversaries can also use a special I/O port to send commands and receive output. Hardware checks, such as the presence of the fan, temperature, and audio devices, could also be used to gather evidence that can be indicative a virtual environment. Adversaries may also query for specific readings from these devices.[3]

Adversaries may modify system firmware to persist on systems.The BIOS (Basic Input/Output System) and The Unified Extensible Firmware Interface (UEFI) or Extensible Firmware Interface (EFI) are examples of system firmware that operate as the software interface between the operating system and hardware of a computer.[1][2][3]

System firmware like BIOS and (U)EFI underly the functionality of a computer and may be modified by an adversary to perform or assist in malicious activity. Capabilities exist to overwrite the system firmware, which may give sophisticated adversaries a means to install malicious firmware updates as a means of persistence on a system that may be difficult to detect.

Adversaries may attempt to gather information about the system language of a victim in order to infer the geographical location of that host. This information may be used to shape follow-on behaviors, including whether the adversary infects the target and/or attempts specific actions. This decision may be employed by malware developers and operators to reduce their risk of attracting the attention of specific law enforcement agencies or prosecution/scrutiny from other entities.[1]

There are various sources of data an adversary could use to infer system language, such as system defaults and keyboard layouts. Specific checks will vary based on the target and/or adversary, but may involve behaviors such as Query Registry and calls to Native API functions.[2]

For example, on a Windows system adversaries may attempt to infer the language of a system by querying the registry key HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Nls\Language or parsing the outputs of Windows API functions GetUserDefaultUILanguage, GetSystemDefaultUILanguage, GetKeyboardLayoutList and GetUserDefaultLangID.[3][4][5]

On a macOS or Linux system, adversaries may query locale to retrieve the value of the $LANG environment variable.

Adversaries may abuse systemctl to execute commands or programs. Systemctl is the primary interface for systemd, the Linux init system and service manager. Typically invoked from a shell, Systemctl can also be integrated into scripts or applications.

Adversaries may use systemctl to execute commands or programs as Systemd Services. Common subcommands include: `systemctl start`, `systemctl stop`, `systemctl enable`, `systemctl disable`, and `systemctl status`.[1]

Adversaries may create or modify systemd services to repeatedly execute malicious payloads as part of persistence. Systemd is a system and service manager commonly used for managing background daemon processes (also known as services) and other system resources.[1] Systemd is the default initialization (init) system on many Linux distributions replacing legacy init systems, including SysVinit and Upstart, while remaining backwards compatible.

Systemd utilizes unit configuration files with the `.service` file extension to encode information about a service's process. By default, system level unit files are stored in the `/systemd/system` directory of the root owned directories (`/`). User level unit files are stored in the `/systemd/user` directories of the user owned directories (`$HOME`).[2]

Inside the `.service` unit files, the following directives are used to execute commands:[3]

* `ExecStart`, `ExecStartPre`, and `ExecStartPost` directives execute when a service is started manually by `systemctl` or on system start if the service is set to automatically start. * `ExecReload` directive executes when a service restarts. * `ExecStop`, `ExecStopPre`, and `ExecStopPost` directives execute when a service is stopped.

Adversaries have created new service files, altered the commands a `.service` file’s directive executes, and modified the user directive a `.service` file executes as, which could result in privilege escalation. Adversaries may also place symbolic links in these directories, enabling systemd to find these payloads regardless of where they reside on the filesystem.[4][5][6]

The `.service` file’s User directive can be used to run service as a specific user, which could result in privilege escalation based on specific user/group permissions.

Systemd services can be created via systemd generators, which support the dynamic generation of unit files. Systemd generators are small executables that run during boot or configuration reloads to dynamically create or modify systemd unit files by converting non-native configurations into services, symlinks, or drop-ins (i.e., Boot or Logon Initialization Scripts).[7][8]

Adversaries may abuse systemd timers to perform task scheduling for initial or recurring execution of malicious code. Systemd timers are unit files with file extension .timer that control services. Timers can be set to run on a calendar event or after a time span relative to a starting point. They can be used as an alternative to Cron in Linux environments.[1] Systemd timers may be activated remotely via the systemctl command line utility, which operates over SSH.[2]

Each .timer file must have a corresponding .service file with the same name, e.g., example.timer and example.service. .service files are Systemd Service unit files that are managed by the systemd system and service manager.[3] Privileged timers are written to /etc/systemd/system/ and /usr/lib/systemd/system while user level are written to ~/.config/systemd/user/.

An adversary may use systemd timers to execute malicious code at system startup or on a scheduled basis for persistence.[4][5][6] Timers installed using privileged paths may be used to maintain root level persistence. Adversaries may also install user level timers to achieve user level persistence.[7]

Adversaries can manipulate or abuse the Transparency, Consent, & Control (TCC) service or database to grant malicious executables elevated permissions. TCC is a Privacy & Security macOS control mechanism used to determine if the running process has permission to access the data or services protected by TCC, such as screen sharing, camera, microphone, or Full Disk Access (FDA).

When an application requests to access data or a service protected by TCC, the TCC daemon (`tccd`) checks the TCC database, located at `/Library/Application Support/com.apple.TCC/TCC.db` (and `~/` equivalent), and an overwrites file (if connected to an MDM) for existing permissions. If permissions do not exist, then the user is prompted to grant permission. Once permissions are granted, the database stores the application's permissions and will not prompt the user again unless reset. For example, when a web browser requests permissions to the user's webcam, once granted the web browser may not explicitly prompt the user again.[1]

Adversaries may access restricted data or services protected by TCC through abusing applications previously granted permissions through Process Injection or executing a malicious binary using another application. For example, adversaries can use Finder, a macOS native app with FDA permissions, to execute a malicious AppleScript. When executing under the Finder App, the malicious AppleScript inherits access to all files on the system without requiring a user prompt. When System Integrity Protection (SIP) is disabled, TCC protections are also disabled. For a system without SIP enabled, adversaries can manipulate the TCC database to add permissions to their malicious executable through loading an adversary controlled TCC database using environment variables and Launchctl.[2][3]

Adversaries may abuse netbooting to load an unauthorized network device operating system from a Trivial File Transfer Protocol (TFTP) server. TFTP boot (netbooting) is commonly used by network administrators to load configuration-controlled network device images from a centralized management server. Netbooting is one option in the boot sequence and can be used to centralize, manage, and control device images.

Adversaries may manipulate the configuration on the network device specifying use of a malicious TFTP server, which may be used in conjunction with Modify System Image to load a modified image on device startup or reset. The unauthorized image allows adversaries to modify device configuration, add malicious capabilities to the device, and introduce backdoors to maintain control of the network device while minimizing detection through use of a standard functionality. This technique is similar to ROMMONkit and may result in the network device running a modified image. [1]

Adversaries may abuse permission configurations that allow them to gain temporarily elevated access to cloud resources. Many cloud environments allow administrators to grant user or service accounts permission to request just-in-time access to roles, impersonate other accounts, pass roles onto resources and services, or otherwise gain short-term access to a set of privileges that may be distinct from their own.

Just-in-time access is a mechanism for granting additional roles to cloud accounts in a granular, temporary manner. This allows accounts to operate with only the permissions they need on a daily basis, and to request additional permissions as necessary. Sometimes just-in-time access requests are configured to require manual approval, while other times the desired permissions are automatically granted.[1]

Account impersonation allows user or service accounts to temporarily act with the permissions of another account. For example, in GCP users with the `iam.serviceAccountTokenCreator` role can create temporary access tokens or sign arbitrary payloads with the permissions of a service account, while service accounts with domain-wide delegation permission are permitted to impersonate Google Workspace accounts.[2][3][4][5] In Exchange Online, the `ApplicationImpersonation` role allows a service account to use the permissions associated with specified user accounts.[6]

Many cloud environments also include mechanisms for users to pass roles to resources that allow them to perform tasks and authenticate to other services. While the user that creates the resource does not directly assume the role they pass to it, they may still be able to take advantage of the role's access -- for example, by configuring the resource to perform certain actions with the permissions it has been granted. In AWS, users with the `PassRole` permission can allow a service they create to assume a given role, while in GCP, users with the `iam.serviceAccountUser` role can attach a service account to a resource.[7][2]

While users require specific role assignments in order to use any of these features, cloud administrators may misconfigure permissions. This could result in escalation paths that allow adversaries to gain access to resources beyond what was originally intended.[8][9]

**Note:** this technique is distinct from Additional Cloud Roles, which involves assigning permanent roles to accounts rather than abusing existing permissions structures to gain temporarily elevated access to resources. However, adversaries that compromise a sufficiently privileged account may grant another account they control Additional Cloud Roles that would allow them to also abuse these features. This may also allow for greater stealth than would be had by directly using the highly privileged account, especially when logs do not clarify when role impersonation is taking place.[10]

Adversaries may abuse components of Terminal Services to enable persistent access to systems. Microsoft Terminal Services, renamed to Remote Desktop Services in some Windows Server OSs as of 2022, enable remote terminal connections to hosts. Terminal Services allows servers to transmit a full, interactive, graphical user interface to clients via RDP.[1]

Windows Services that are run as a "generic" process (ex: svchost.exe) load the service's DLL file, the location of which is stored in a Registry entry named ServiceDll.[2] The termsrv.dll file, typically stored in `%SystemRoot%\System32\`, is the default ServiceDll value for Terminal Services in `HKLM\System\CurrentControlSet\services\TermService\Parameters\`.

Adversaries may modify and/or replace the Terminal Services DLL to enable persistent access to victimized hosts.[3] Modifications to this DLL could be done to execute arbitrary payloads (while also potentially preserving normal termsrv.dll functionality) as well as to simply enable abusable features of Terminal Services. For example, an adversary may enable features such as concurrent Remote Desktop Protocol sessions by either patching the termsrv.dll file or modifying the ServiceDll value to point to a DLL that provides increased RDP functionality.[4][5] On a non-server Windows OS this increased functionality may also enable an adversary to avoid Terminal Services prompts that warn/log out users of a system when a new RDP session is created.

Adversaries may inject malicious code into hijacked processes in order to evade process-based defenses as well as possibly elevate privileges. Thread Execution Hijacking is a method of executing arbitrary code in the address space of a separate live process.

Thread Execution Hijacking is commonly performed by suspending an existing process then unmapping/hollowing its memory, which can then be replaced with malicious code or the path to a DLL. A handle to an existing victim process is first created with native Windows API calls such as OpenThread. At this point the process can be suspended then written to, realigned to the injected code, and resumed via SuspendThread , VirtualAllocEx, WriteProcessMemory, SetThreadContext, then ResumeThread respectively.[1]

This is very similar to Process Hollowing but targets an existing process rather than creating a process in a suspended state.

Running code in the context of another process may allow access to the process's memory, system/network resources, and possibly elevated privileges. Execution via Thread Execution Hijacking may also evade detection from security products since the execution is masked under a legitimate process.

Adversaries may inject malicious code into processes via thread local storage (TLS) callbacks in order to evade process-based defenses as well as possibly elevate privileges. TLS callback injection is a method of executing arbitrary code in the address space of a separate live process.

TLS callback injection involves manipulating pointers inside a portable executable (PE) to redirect a process to malicious code before reaching the code's legitimate entry point. TLS callbacks are normally used by the OS to setup and/or cleanup data used by threads. Manipulating TLS callbacks may be performed by allocating and writing to specific offsets within a process’ memory space using other Process Injection techniques such as Process Hollowing.[1]

Running code in the context of another process may allow access to the process's memory, system/network resources, and possibly elevated privileges. Execution via TLS callback injection may also evade detection from security products since the execution is masked under a legitimate process.

Adversaries may search private data from threat intelligence vendors for information that can be used during targeting. Threat intelligence vendors may offer paid feeds or portals that offer more data than what is publicly reported. Although sensitive details (such as customer names and other identifiers) may be redacted, this information may contain trends regarding breaches such as target industries, attribution claims, and successful TTPs/countermeasures.[1]

Adversaries may search in private threat intelligence vendor data to gather actionable information. If a threat actor is searching for information on their own activities, that falls under Search Threat Vendor Data. Information reported by vendors may also reveal opportunities other forms of reconnaissance (ex: Search Open Websites/Domains), establishing operational resources (ex: Develop Capabilities or Obtain Capabilities), and/or initial access (ex: Exploit Public-Facing Application or External Remote Services).

Adversaries may employ various time-based methods to detect virtualization and analysis environments, particularly those that attempt to manipulate time mechanisms to simulate longer elapses of time. This may include enumerating time-based properties, such as uptime or the system clock.

Adversaries may use calls like `GetTickCount` and `GetSystemTimeAsFileTime` to discover if they are operating within a virtual machine or sandbox, or may be able to identify a sandbox accelerating time by sampling and calculating the expected value for an environment's timestamp before and after execution of a sleep function.[1]

Adversaries may abuse time providers to execute DLLs when the system boots. The Windows Time service (W32Time) enables time synchronization across and within domains.[1] W32Time time providers are responsible for retrieving time stamps from hardware/network resources and outputting these values to other network clients.[2]

Time providers are implemented as dynamic-link libraries (DLLs) that are registered in the subkeys of `HKEY_LOCAL_MACHINE\System\CurrentControlSet\Services\W32Time\TimeProviders\`.[2] The time provider manager, directed by the service control manager, loads and starts time providers listed and enabled under this key at system startup and/or whenever parameters are changed.[2]

Adversaries may abuse this architecture to establish persistence, specifically by creating a new arbitrarily named subkey pointing to a malicious DLL in the `DllName` value. Administrator privileges are required for time provider registration, though execution will run in context of the Local Service account.[3]

Adversaries may modify file time attributes to hide new files or changes to existing files. Timestomping is a technique that modifies the timestamps of a file (the modify, access, create, and change times), often to mimic files that are in the same folder and blend malicious files with legitimate files.

In Windows systems, both the `$STANDARD_INFORMATION` (`$SI`) and `$FILE_NAME` (`$FN`) attributes record times in a Master File Table (MFT) file.[1] `$SI` (dates/time stamps) is displayed to the end user, including in the File System view, while `$FN` is dealt with by the kernel.[2]

Modifying the `$SI` attribute is the most common method of timestomping because it can be modified at the user level using API calls. `$FN` timestomping, however, typically requires interacting with the system kernel or moving or renaming a file.[1]

Adversaries modify timestamps on files so that they do not appear conspicuous to forensic investigators or file analysis tools. In order to evade detections that rely on identifying discrepancies between the `$SI` and `$FN` attributes, adversaries may also engage in “double timestomping” by modifying times on both attributes simultaneously.[3]

In Linux systems and on ESXi servers, threat actors may attempt to perform timestomping using commands such as `touch -a -m -t ` (which sets access and modification times to a specific value) or `touch -r ` (which sets access and modification times to match those of another file).[4][5]

Timestomping may be used along with file name Masquerading to hide malware and tools.[6]

Adversaries may duplicate then impersonate another user's existing token to escalate privileges and bypass access controls. For example, an adversary can duplicate an existing token using `DuplicateToken` or `DuplicateTokenEx`.[1] The token can then be used with `ImpersonateLoggedOnUser` to allow the calling thread to impersonate a logged on user's security context, or with `SetThreadToken` to assign the impersonated token to a thread.

An adversary may perform Token Impersonation/Theft when they have a specific, existing process they want to assign the duplicated token to. For example, this may be useful for when the target user has a non-network logon session on the system.

When an adversary would instead use a duplicated token to create a new process rather than attaching to an existing process, they can additionally Create Process with Token using `CreateProcessWithTokenW` or `CreateProcessAsUserW`. Token Impersonation/Theft is also distinct from Make and Impersonate Token in that it refers to duplicating an existing token, rather than creating a new one.