Hardening Microsoft Windows 10 version 21H1 Workstations

Introduction

Workstations are often targeted by an adversary using malicious websites, emails or removable media in an attempt to extract sensitive information. Hardening workstations is an important part of reducing this risk.

This publication provides recommendations on hardening workstations using Enterprise and Education editions of Microsoft Windows 10 version 21H1. Before implementing recommendations in this publication, thorough testing should be undertaken to ensure the potential for unintended negative impacts on business processes is reduced as much as possible.

While this publication refers to workstations, most recommendations are equally applicable to servers (with the exception of Domain Controllers) using Microsoft Windows Server version 21H1 or Microsoft Windows Server 2019.

Security features discussed in this publication, along with the names and locations of Group Policy settings, are taken from Microsoft Windows 10 version 21H1 – some differences will exist for earlier versions of Microsoft Windows 10.

For cloud-based device managers, such as Microsoft Endpoint Manager, equivalents can be found for many of the Group Policy settings. Alternatively, there is often a function to import Group Policy settings into cloud-based device managers.

A summary of the changes from the previous release of this publication are:

• exceptions for default application control rulesets were updated

• privilege escalation guidance was updated to automatically deny elevation requests for standard users

• guidance on Chromium-based Microsoft Edge was added

• guidance on Windows Hello for Business was added

• guidance on Windows Update for Business was added

• guidance on Windows To Go was removed.

High priorities

The following recommendations, listed in alphabetical order, should be treated as high priorities when hardening Microsoft Windows 10 workstations.

Application hardening

When applications are installed they are often not pre-configured in a secure state. By default, many applications enable functionality that isn’t required by any users while in-built security functionality may be disabled or set at a lower security level. For example, Microsoft Office by default allows untrusted macros in Office documents to automatically execute without user interaction. To reduce this risk, applications should have any in-built security functionality enabled and appropriately configured along with unrequired functionality disabled. This is especially important for key applications such as office productivity suites (e.g. Microsoft Office), PDF readers (e.g. Adobe Reader), web browsers (e.g. Microsoft Internet Explorer, Mozilla Firefox or Google Chrome), common web browser plugins (e.g. Adobe Flash), email clients (Microsoft Outlook) and software platforms (e.g. Oracle Java Platform and Microsoft .NET Framework). In addition, vendors may provide guidance on configuring their products securely. For example, Microsoft provides security baselines for their products on their Microsoft Security Baseline Blog. In such cases, vendor guidance should be followed to assist in securely configuring their products.

The Australian Cyber Security Centre also provides guidance for hardening Microsoft Office. For more information see the Hardening Microsoft 365, Office 2021, Office 2019 and Office 2016 publication.

Application versions and patches

While some vendors may release new application versions to address security vulnerabilities, others may release patches. If new application versions and patches for applications are not installed it can allow an adversary to easily compromise workstations. This is especially important for key applications that interact with content from untrusted sources such as office productivity suites (e.g. Microsoft Office), PDF readers (e.g. Adobe Reader), web browsers (e.g. Microsoft Internet Explorer, Mozilla Firefox or Google Chrome), common web browser plugins (e.g. Adobe Flash), email clients (Microsoft Outlook) and software platforms (e.g. Oracle Java Platform and Microsoft .NET Framework). To reduce this risk, new application versions and patches for applications should be applied in an appropriate timeframe as determined by the severity of security vulnerabilities they address and any mitigating measures already in place. In cases where a previous version of an application continues to receive support in the form of patches, it still should be upgraded to the latest version to receive the benefit of any new security functionality.

For more information on determining the severity of security vulnerabilities and timeframes for applying new application versions and patches for applications see the Assessing Security Vulnerabilities and Applying Patches publication.

Application control

An adversary can email malicious code, or host malicious code on a compromised website, and use social engineering techniques to convince users into executing it. Such malicious code often aims to exploit security vulnerabilities in existing applications and does not need to be installed to be successful. Application control can be an extremely effective mechanism in not only preventing malicious code from executing, but also ensuring only approved applications can be installed.

When developing application control rules, starting from scratch is a more secure method than relying on a list of executable content currently residing on a workstation. Furthermore, it is preferable that organisations define their own application control ruleset rather than relying on rulesets from application control vendors. This application control ruleset should then be regularly assessed to determine if it remains fit for purpose.

For more information on application control and how it can be appropriately implemented see the Implementing Application Control publication.

Attack Surface Reduction

Attack Surface Reduction (ASR), a security feature of Microsoft Windows 10, forms part of Microsoft Defender Exploit Guard. It is designed to combat the threat of malware exploiting legitimate functionality in Microsoft Office applications. In order to use ASR, Microsoft Defender Antivirus must be configured as the primary real-time antivirus scanning engine on workstations.

ASR offers a number of attack surface reduction rules, these include:

• Block executable content from email client and webmail
  BE9BA2D9-53EA-4CDC-84E5-9B1EEEE46550

• Block all Office applications from creating child processes
  D4F940AB-401B-4EFC-AADC-AD5F3C50688A

• Block Office applications from creating executable content
  3B576869-A4EC-4529-8536-B80A7769E899

• Block Office applications from injecting code into other processes
  75668C1F-73B5-4CF0-BB93-3ECF5CB7CC84

• Block JavaScript or VBScript from launching downloaded executable content
  D3E037E1-3EB8-44C8-A917-57927947596D

• Block execution of potentially obfuscated scripts
  5BEB7EFE-FD9A-4556-801D-275E5FFC04CC

• Block Win32 API calls from Office macro
  92E97FA1-2EDF-4476-BDD6-9DD0B4DDDC7B

• Block executable files from running unless they meet a prevalence, age, or trusted list criterion
  01443614-CD74-433A-B99E-2ECDC07BFC25

• Use advanced protection against ransomware
  C1DB55AB-C21A-4637-BB3F-A12568109D35

• Block credential stealing from the Windows local security authority subsystem (lsass.exe)
  9E6C4E1F-7D60-472F-BA1A-A39EF669E4B2

• Block process creations originating from PSExec and WMI commands
  D1E49AAC-8F56-4280-B9BA-993A6D77406C

• Block untrusted and unsigned processes that run from USB
  B2B3F03D-6A65-4F7B-A9C7-1C7EF74A9BA4

• Block Office communication application from creating child processes
  26190899-1602-49E8-8B27-EB1D0A1CE869

• Block Adobe Reader from creating child processes
  7674BA52-37EB-4A4F-A9A1-F0F9A1619A2C

• Block persistence through WMI event subscription
  E6DB77E5-3DF2-4CF1-B95A-636979351E5B.

Organisations should either implement ASR using Microsoft Defender Antivirus or use third party antivirus solutions that offer similar functionality to those provided by ASR. For older versions of Microsoft Windows, alternative measures will need to be implemented to mitigate certain threats addressed by ASR, such as the likes of Dynamic Data Exchange (DDE) attacks.

For organisations using Microsoft Defender Antivirus, the following Group Policy settings can be implemented to enforce the above ASR rules.

Credential caching

Cached credentials are stored in the Security Accounts Manager (SAM) database and can allow a user to log onto a workstation they have previously logged onto even if the domain is not available. Whilst this functionality may be desirable from an availability of services perspective, this functionality can be abused by an adversary who can retrieve these cached credentials (potentially Domain Administrator credentials in a worst-case scenario). To reduce this risk, cached credentials should be limited to only one previous logon.

The following Group Policy settings can be implemented to disable credential caching.

Within an active user session, credentials are cached within the Local Security Authority Subsystem Service (LSASS) process (including the user’s passphrase in plaintext if WDigest authentication is enabled) to allow for access to network resources without users having to continually enter their credentials. Unfortunately, these credentials are at risk of theft by an adversary. To reduce this risk, WDigest authentication should be disabled.

Windows Defender Credential Guard, a security feature of Microsoft Windows 10, is also designed to assist in protecting the LSASS process.

The following Group Policy settings can be implemented to disable WDigest authentication and enable Credential Guard functionality, assuming all software, firmware and hardware prerequisites are met. Note, the MS Security Guide Group Policy settings are available as part of the Microsoft Security Compliance Toolkit.

Controlled Folder Access

Controlled Folder Access, a security feature of Microsoft Windows 10, forms part of Microsoft Defender Exploit Guard. It is designed to combat the threat of ransomware.

In order to use Controlled Folder Access, Microsoft Defender Antivirus must be configured as the primary real-time antivirus scanning engine on workstations. Other third party antivirus solutions may offer similar functionality as part of their offerings.

The following Group Policy settings can be implemented to implement Controlled Folder Access.

Credential entry

When users enter their credentials on a workstation it provides an opportunity for malicious code, such as a key logging application, to capture the credentials. To reduce this risk, users should be authenticated by using a trusted path to enter their credentials on the Secure Desktop.

The following Group Policy settings can be implemented to ensure credentials are entered in a secure manner.

Early Launch Antimalware

Another key security feature of Trusted Boot, supported by Microsoft Windows 10 and motherboards with an Unified Extensible Firmware Interface (UEFI), is Early Launch Antimalware (ELAM). Used in conjunction with Secure Boot, an ELAM driver can be registered as the first non-Microsoft driver that will be initialised on a workstation as part of the boot process, thus allowing it to verify all subsequent drivers before they are initialised. The ELAM driver is capable of allowing only known good drivers to initialise; known good and unknown drivers to initialise; known good, unknown and bad but critical drivers to initialise; or all drivers to initialise. To reduce the risk of malicious drivers, only known good and unknown drivers should be allowed to be initialised during the boot process.

The following Group Policy setting can be implemented to ensure only known good and unknown drivers will be initialised at boot time.

Elevating privileges

Microsoft Windows provides the ability to require confirmation from users, via the User Access Control (UAC) functionality, before any sensitive actions are performed. The default settings allow privileged users to perform sensitive actions without first providing credentials and while standard users must provide privileged credentials they are not required to do so via a trusted path on the Secure Desktop. This provides an opportunity for an adversary that gains access to an open session of a privileged user to perform sensitive actions at will or for malicious code to capture any credentials entered via a standard user when attempting to elevate their privileges. To reduce this risk, UAC functionality should be implemented to ensure all sensitive actions are authorised by providing credentials on the Secure Desktop.

The following Group Policy settings can be implemented to configure UAC functionality effectively.

Exploit protection

An adversary that develops exploits for Microsoft Windows or third party applications will have a higher success rate when security measures designed by Microsoft to help prevent security vulnerabilities from being exploited are not implemented. Microsoft Defender’s exploit protection functionality, a security feature of Microsoft Windows 10, provides system-wide and application-specific security measures. Exploit protection is designed to replace the Enhanced Mitigation Experience Toolkit (EMET) that was used on earlier versions of Microsoft Windows 10.

System-wide security measures configurable via exploit protection include: Control Flow Guard (CFG), Data Execution Prevention (DEP), mandatory Address Space Layout Randomization (ASLR), bottom-up ASLR, Structured Exception Handling Overwrite Protection (SEHOP) and heap corruption protection.

Many more application-specific security measures are also available, however, they will require testing (either within a test environment or using audit mode) beforehand to limit the likelihood of any unintended consequences. As such, a staged approach to implementing application-specific security measures is prudent. In doing so, applications that ingest arbitrary untrusted data from the internet should be prioritised.

The following Group Policy settings can be implemented to define exploit protection settings and to prevent users from modifying these settings on their devices.

In addition, the following Group Policy setting can be implemented to ensure DEP is not disabled for File Explorer.

Furthermore, the following Group Policy setting can be implemented to force the use of SEHOP. Note, the MS Security Guide Group Policy settings are available as part of the Microsoft Security Compliance Toolkit.

Local administrator accounts

When built-in administrator accounts are used with common account names and passwords it can allow an adversary that compromises these credentials on one workstation to easily transfer across the network to other workstations. Even if built-in administrator accounts are uniquely named and have unique passwords, an adversary can still identify these accounts based on their security identifier (i.e. S-1-5-21-domain-500) and use this information to focus any attempts to brute force credentials on a workstation if they can get access to the SAM database. To reduce this risk, built-in administrator accounts should be disabled. Instead, domain accounts with local administrative privileges, but without domain administrative privileges, should be used for workstation management.

The following Group Policy setting can be implemented to disable built-in administrator accounts.

If a common local administrator account absolutely must be used for workstation management then Microsoft’s Local Administrator Password Solution (LAPS) needs to be used to ensure unique passphrases are used for each workstation. In addition, User Account Control restrictions should be applied to remote connections using such accounts. Note, the MS Security Guide Group Policy settings are available as part of the Microsoft Security Compliance Toolkit.

Measured Boot

The third key security feature of Trusted Boot, supported by Microsoft Windows 10 and motherboards with both an UEFI and a Trusted Platform Module (TPM), is Measured Boot. Measured Boot is used to develop a reliable log of components that are initialised before the ELAM driver. This information can then be scrutinised by antimalware software for signs of tampering of boot components. To reduce the risk that malicious changes to boot components go unnoticed, Measured Boot should be used on workstations that support it.

Microsoft Edge

Microsoft Edge is a web browser that was first introduced in Microsoft Windows 10 to replace Internet Explorer 11. Microsoft Edge contains significant security enhancements (the most recent version being based on the Chromium project) over Internet Explorer 11 and should be used wherever possible. Furthermore, as Microsoft Edge contains an ‘IE mode’, Internet Explorer 11 should be disabled or removed from Microsoft Windows 10 to reduce the operating system’s attack surface.

For organisations using Microsoft Edge instead of third party web browsers, the following Group Policy settings (once the supporting Group Policy Administrative Templates have been installed) can be implemented to harden Microsoft Edge, including Microsoft Defender SmartScreen. Note, the following Group Policy settings are taken from Microsoft Edge version 90, some differences may exist in different versions.

For organisations that have yet to disable or remove Internet Explorer 11 and switch to using ‘IE mode’ within Microsoft Edge, only specified intranet websites should be accessible from Internet Explorer 11 with all other websites being directed to open within Microsoft Edge. This can be achieved using the following Group Policy settings.

Multi-factor authentication

As privileged credentials often allow users to bypass security functionality put in place to protect workstations, and are susceptible to key logging applications, it is important that they are appropriately protected against compromise. In addition, an adversary that brute forces captured password hashes can gain access to workstations if multi-factor authentication hasn’t been implemented. To reduce this risk, hardware-based multi-factor authentication should be used for privileged users, remote access and any access to important or sensitive data repositories.

Organisations may consider whether Windows Hello for Business (WHfB) is suitable for their environment. Notably, WHfB can be configured with a personal identification number (PIN) or face/fingerprint recognition to unlock the use of asymmetric cryptography stored in a TPM in order to authenticate users. Note, the use of TPMs places additional importance on patching TPMs for security vulnerabilities and decommissioning those devices that are not able to be patched. Organisations may also choose to enforce the use of the latest versions of TPMs when using WHfB. Finally, Microsoft has issued guidance on the use of FIDO2 security tokens as part of multi-factor authentication for Microsoft Windows logons.

For more information on how to effectively implement multi-factor authentication see the Implementing Multi-Factor Authentication publication.

Operating system architecture

The x64 (64-bit) versions of Microsoft Windows include additional security functionality that the x86 (32-bit) versions lack. This includes native hardware-based Data Execution Prevention (DEP) kernel support, Kernel Patch Protection (PatchGuard), mandatory device driver signing and lack of support for malicious 32-bit drivers. Using x86 (32-bit) versions of Microsoft Windows exposes organisations to exploit techniques mitigated by x64 (64-bit) versions of Microsoft Windows. To reduce this risk, workstations should use the x64 (64-bit) versions of Microsoft Windows.

Operating system patching

Patches are released either in response to previously disclosed security vulnerabilities or to proactively address security vulnerabilities that have not yet been publicly disclosed. In the case of disclosed security vulnerabilities, it is possible that exploits have already been developed and are freely available in common hacking tools. In the case of patches for security vulnerabilities that have not yet been publically disclosed, it is relatively easy for an adversary to use freely available tools to identify the security vulnerability being patched and develop an associated exploit. This activity can be undertaken in less than one day and has led to an increase in 1-day attacks. To reduce this risk, operating system patches and driver updates should be centrally managed, deployed and applied in an appropriate timeframe as determined by the severity of the security vulnerability and any mitigating measures already in place.

Previously, operating system patching was typically achieved by using Microsoft Endpoint Configuration Manager, or Microsoft Windows Server Update Services (WSUS), along with Wake-on-LAN functionality to facilitate patching outside of core business hours. However, Windows Update for Business may replace or supplement many WSUS functions. Configuration of Windows Update for Business can be applied through Group Policy settings or the equivalent settings in Microsoft Endpoint Manager. Microsoft has also issued guidance on common misconfigurations relating to Windows updates.

For more information on determining the severity of security vulnerabilities and timeframes for applying patches see the Assessing Security Vulnerabilities and Applying Patches publication.

The following Group Policy settings can be implemented to ensure operating systems remain appropriately patched.

Furthermore, if a Windows Server Update Services (WSUS) server is used, the following Group Policy setting can be implemented.

Operating system version

Microsoft Windows 10 has introduced improvements in security functionality over previous versions of Microsoft Windows. This has made it more difficult for an adversary to craft reliable exploits for security vulnerabilities they discovered. Using older versions of Microsoft Windows, including previous versions of Microsoft Windows 10, exposes organisations to exploit techniques that have since been mitigated in newer versions of Microsoft Windows. To reduce this risk, workstations should use the latest version of Microsoft Windows 10.

Restricting privileged accounts

Providing users with a privileged account for day to day usage poses a risk that they will use this account for external web and email access. This is of particular concern as privileged users have the ability to execute malicious code with privileged access rather than standard access. To reduce this risk, users that don’t require privileged access should not be granted privileged accounts while users that require privileged access should have separate standard and privileged accounts with different credentials. In addition, any privileged accounts used should have external web and email access blocked.

For more information on the use of privileged accounts and minimising their usage see the Restricting Administrative Privileges publication.

Secure Boot

Another method for malicious code to maintain persistence and prevent detection is to replace the default boot loader for Microsoft Windows with a malicious version. In such cases the malicious boot loader executes at boot time and loads Microsoft Windows without any indication that it is present. Such malicious boot loaders are extremely difficult to detect and can be used to conceal malicious code on workstations. To reduce this risk, motherboards with Secure Boot functionality should be used. Secure Boot, a component of Trusted Boot, is a security feature of Microsoft Windows 10 and motherboards with an UEFI. Secure Boot works by checking at boot time that the boot loader is signed and matches a Microsoft signed certificate stored in the UEFI. If the certificate signatures match the boot loader is allowed to run, otherwise it is prevented from running and the workstation will not boot.

Medium priorities

The following recommendations, listed in alphabetical order, should be treated as medium priorities when hardening Microsoft Windows 10 workstations.

Account lockout policy

Allowing unlimited attempts to access workstations will fail to prevent an adversary’s attempts to brute force authentication measures. To reduce this risk, accounts should be locked out after a defined number of invalid authentication attempts. The threshold for locking out accounts does not need to be overly restrictive in order to be effective. For example, a threshold of 5 incorrect attempts, with a reset period of 15 minutes for the lockout counter, will prevent any brute force attempt while being unlikely to lock out a legitimate user who accidently enters their password incorrectly a few times.

The following Group Policy settings can be implemented to achieve a reasonable lockout policy.

Anonymous connections

An adversary can use anonymous connections to gather information about the state of workstations. Information that can be gathered from anonymous connections (i.e. using the net use command to connect to the IPC$ share) can include lists of users and groups, SIDs for accounts, lists of shares, workstation policies, operating system versions and patch levels. To reduce this risk, anonymous connections to workstations should be disabled.

The following Group Policy settings can be implemented to disable the use of anonymous connections.

Antivirus software

An adversary can develop malicious code to exploit security vulnerabilities in software not detected and remedied by vendors during testing. As significant time and effort is often involved in the development of functioning and reliable exploits, an adversary will often reuse their exploits as much as possible before being forced to develop new exploits. To reduce this risk, endpoint security applications with signature-based antivirus functionality should be implemented. In doing so, signatures should be updated at least on a daily basis.

Whilst using signature-based antivirus functionality can assist in reducing risk, they are only effective when a particular piece of malicious code has already been profiled and signatures are current. An adversary can create variants of known malicious code, or develop new unseen malicious code, to bypass traditional signature-based detection mechanisms. To reduce this risk, endpoint security applications with host-based intrusion prevention functionality, or equivalent functionality leveraging cloud-based services, should also be implemented. In doing so, such functionality should be set at the highest level available.

If using Microsoft Defender Antivirus, the following Group Policy settings can be implemented to optimally configure it.

Attachment Manager

The Attachment Manager within Microsoft Windows works in conjunction with applications such as the Microsoft Office suite and Internet Explorer to help protect workstations from attachments that have been received via email or downloaded from the internet. The Attachment Manager classifies files as high, medium or low risk based on the zone they originated from and the type of file. Based on the risk to the workstation, the Attachment Manager will either issue a warning to a user or prevent them from opening a file. If zone information is not preserved, or can be removed, it can allow an adversary to socially engineer a user to bypass protections afforded by the Attachment Manager. To reduce this risk, the Attachment Manager should be configured to preserve and protect zone information for files.

The following Group Policy settings can be implemented to ensure zone information associated with attachments is preserved and protected.

Audit event management

Failure to capture and analyse security related audit events from workstations can result in intrusions going unnoticed. In addition, the lack of such information can significantly hamper investigations following a security incident. To reduce this risk, security related audit events from workstations should be captured and routinely analysed.

The following Group Policy settings can be implemented to ensure security related audit events are appropriately captured.

Furthermore, the following Group Policy settings can be implemented to enable a comprehensive auditing strategy.

Autoplay and AutoRun

When enabled, Autoplay will automatically begin reading from a drive or media source as soon as it is used with a workstation, while AutoRun commands, generally in an autorun.inf file on the media, can be used to automatically execute any file on the media without user interaction. This functionality can be exploited by an adversary to automatically execute malicious code. To reduce this risk, Autoplay and AutoRun functionality should be disabled.

The following Group Policy settings can be implemented to disable Autoplay and AutoRun functionality.

BIOS and UEFI passwords

An adversary with access to a workstation’s Basic Input/Output System (BIOS) or UEFI can modify the hardware configuration of the workstation to introduce attack vectors or weaken security functionality within the workstation’s operating system. This can include disabling security functionality in the CPU, modifying allowed boot devices and enabling insecure communications interfaces such as FireWire and Thunderbolt. To reduce this risk, strong BIOS and UEFI passwords should be used for all workstations to prevent unauthorised access.

Boot devices

By default, workstations are often configured to boot from optical media, or even USB media, in preference to hard drives. An adversary with physical access to such workstations can boot from their own media in order to gain access to the content of the hard drives. With this access, an adversary can reset local user account passwords or gain access to the local SAM database to steal password hashes for offline brute force cracking attempts. To reduce this risk, workstations should be restricted to only booting from the designated primary system drive.

Bridging networks

When workstations have multiple network interfaces, such as an Ethernet interface and a wireless interface, it is possible to establish a bridge between the connected networks. For example, when using an Ethernet interface to connect to an organisation’s wired network and a wireless interface to connect to another non-organisation controlled network such as a public wireless hotspot. When bridges are created between such networks an adversary can directly access the wired network from the wireless network to extract sensitive information. To reduce this risk, the ability to install and configure network bridges between different networks should be disabled. This won’t prevent an adversary from compromising a workstation via the wireless network and then using malicious software as a medium to indirectly access the wired network. This can only be prevented by manually disabling all wireless interfaces when connecting to wired networks.

The following Group Policy settings can be implemented to disable the ability to install and configure network bridges.

Built-in guest accounts

When built-in guest accounts are used, it can allow an adversary to log onto a workstation over the network without first needing to compromise legitimate user credentials. To reduce this risk, built-in guest accounts should be disabled.

The following Group Policy settings can be implemented to disable and rename built-in guest accounts.

CD burner access

If CD burning functionality is enabled, and CD burners are installed in workstations, an adversary may attempt to steal sensitive information by burning it to CD. To reduce this risk, users should not have access to CD burning functionality except when explicitly required.

The following Group Policy setting can be implemented to prevent access to CD burning functionality, although as this Group Policy setting only prevents access to native CD burning functionality in Microsoft Windows, users should also be prevented from installing third party CD burning applications. Alternatively, CD readers can be used in workstations instead of CD burners.

Centralised audit event logging

Storing audit event logs on workstations poses a risk that an adversary could attempt to modify or delete these logs during an intrusion to cover their tracks. In addition, failure to conduct centralised audit event logging will reduce the visibility of audit events across all workstations, prevent the correlation of audit events and increase the complexity of any investigations after security incidents. To reduce this risk, audit event logs from workstations should be transferred to a secure central logging server.

Command Prompt

An adversary who gains access to a workstation can use the Command Prompt to execute in-built Microsoft Windows tools to gather information about the workstation or domain as well as schedule malicious code to execute on other workstations on the network. To reduce this risk, users should not have Command Prompt access or the ability to execute batch files and scripts. Should a legitimate business requirement exist to allow users to execute batch files (e.g. cmd and bat files); run logon, logoff, startup or shutdown batch file scripts; or use Remote Desktop Services, this risk will need to be accepted.

The following Group Policy setting can be implemented to prevent access to the Command Prompt and script processing functionality.

Direct Memory Access

Communications interfaces that use Direct Memory Access (DMA) can allow an adversary with physical access to a workstation to directly access the contents of a workstation’s memory. This can be used to read sensitive contents such as cryptographic keys or to write malicious code directly into memory. To reduce this risk, communications interfaces that allow DMA (e.g. FireWire and Thunderbolt) should be disabled. This can be achieved either physically (e.g. using epoxy) or by using software controls (e.g. disabling the functionality in the BIOS or UEFI; removing the SBP-2 driver and disabling the FireWire and Thunderbolt controllers; or using an end point protection solution).

The following Group Policy settings can be implemented to remove the SBP-2 driver as well as disable the FireWire and Thunderbolt controllers. Note, Intel-based systems have included built-in kernel DMA protection for Thunderbolt 3 by default, however, such protections are not foolproof.

Drive encryption

An adversary with physical access to a workstation may be able to use a bootable CD/DVD or USB media to load their own operating environment. From this environment, they can access the local file system to gain access to sensitive information or the SAM database to access password hashes. In addition, an adversary that gains access to a stolen or unsanitised hard drive, or other removable media, will be to recover its contents when connected to another machine on which they have administrative access and can take ownership of files. To reduce this risk, AES-based full disk encryption should be used to protect the contents of hard drives from unauthorised access. The use of full disk encryption may also contribute to streamlining media sanitisation during decommissioning processes.

If Microsoft BitLocker is used, the following Group Policy settings should be implemented. Key lengths have been aligned with the minimums specified in the Information Security Manual.

Endpoint device control

An adversary with physical access to a workstation may attempt to connect unauthorised USB media or other devices with mass storage functionality (e.g. smartphones, digital music players or cameras) to facilitate malicious code infections or the unauthorised copying of sensitive information. To reduce this risk, endpoint device control functionality should be appropriately implemented to control the use of all removable storage devices.

The following Group Policy setting can be implemented to disable the use of removable storage devices.

Alternatively, if specific classes of removable storage devices are required to meet business requirements, the execute, read and write permissions should be controlled on a class by class basis.

The following Group Policy settings provide a sample implementation that allows data to be read from but not executed from or written to all classes of removable storage devices.

File and print sharing

Users sharing files from their workstations can result in a lack of appropriate access controls being applied to sensitive information and the potential for the propagation of malicious code should file shares have read/write access. To reduce this risk, local file and print sharing should be disabled. Ideally, sensitive information should be centrally managed (e.g. on a network share with appropriate access controls). Disabling file and print sharing will not affect a user’s ability to access shared drives and printers on a network.

The following Group Policy settings can be implemented to prevent users from sharing files.

Group Policy processing

Relying on users to set Group Policy settings for their workstations creates the potential for users to inadvertently misconfigure or disable security functionality without consideration of the impact on the security posture of the workstation. Alternatively, an adversary could exploit this to disable any Local Group Policy settings that are hampering their efforts to extract sensitive information. To reduce this risk, all audit, user rights and security related Group Policy settings should be specified for workstations at an organisational unit or domain level. To ensure these policies aren’t weakened, support for Local Group Policy settings should also be disabled.

The following Group Policy settings can be implemented to ensure only domain-based Group Policy settings are obtained and applied to workstations in a secure manner.

Installing applications and drivers

While the ability to install applications may be a business requirement for users, this privilege can be exploited by an adversary. An adversary can email a malicious application, or host a malicious application on a compromised website, and use social engineering techniques to convince users into installing the application on their workstation. Even if privileged access is required to install applications, users will use their privileged access if they believe, or can be convinced that, the requirement to install the application is legitimate. Additionally, if applications are configured to install using elevated privileges, an adversary can exploit this by creating a Windows Installer installation package to create a new account that belongs to the local built-in administrators group or to install a malicious application. Alternatively, an adversary may attempt to install drivers that are not relevant to a system in order to introduce security vulnerabilities. To reduce this risk, all application and driver installations should be strictly controlled.

The following Group Policy settings can be implemented to control application and driver installations.

Legacy and run once lists

Once malicious code has been copied to a workstation, an adversary with registry access can remotely schedule it to execute (i.e. using the run once list) or to automatically execute each time Microsoft Windows starts (i.e. using the legacy run list). To reduce this risk, legacy and run once lists should be disabled. This may interfere with the operation of legitimate applications that need to automatically execute each time Microsoft Windows starts. In such cases, the Run these programs at user logon Group Policy setting can be used to perform the same function in a more secure manner when defined at a domain level; however, if not used this Group Policy setting should be disabled rather than left in its default undefined state.

The following Group Policy settings can be implemented to disable the use of legacy and run once lists.

Microsoft accounts

A feature of Microsoft Windows 10 is the ability to link Microsoft accounts (formerly Windows Live IDs) to local or domain accounts. When this occurs, a user’s settings and files are stored in the cloud using OneDrive rather than locally or on a domain controller. While this may have the benefit of allowing users to access their settings and files from any workstation (e.g. corporate workstation, home PC, internet cafe) it can also pose a risk to an organisation as they lose control over where sensitive information may be accessed from. To reduce this risk, users should not link Microsoft accounts with local or domain accounts.

The following Group Policy settings can be implemented to disable the ability to link Microsoft accounts to local or domain accounts.

MSS settings

MSS settings are registry values previously identified by Microsoft security experts that can be used for increased security. While many of these registry values are no longer applicable in modern versions of Microsoft Windows, some still provide a security benefit. By failing to specify these MSS settings, an adversary may be able to exploit weaknesses in a workstation’s security posture to gain access to sensitive information. To reduce this risk, MSS settings that are still relevant to modern versions of Microsoft Windows should be specified using Group Policy settings.

The Group Policy Administrative Templates for MSS settings are available from the Microsoft Security Guidance blog. The ADMX and ADML files can be placed in %SystemDrive%\Windows\SYSVOL\domain\Policies\PolicyDefinitions on the Domain Controller and they will automatically be loaded in the Group Policy Management Editor.

The following Group Policy settings can be implemented to configure MSS settings that are still relevant to modern versions of Microsoft Windows.

NetBIOS over TCP/IP

NetBIOS over TCP/IP facilitates a number of intrusion methods. To reduce this risk, NetBIOS over TCP/IP should be disabled. As NetBIOS over TCP/IP is only used to support legacy Microsoft Windows operating systems, such as those prior to Microsoft Windows 2000, there shouldn’t be a business requirement for its use except in very rare circumstances. NetBIOS over TCP/IP can be disabled by setting the NetBIOS settings under the IPv4 WINS settings on each network interface to Disable NetBIOS over TCP/IP. NetBIOS over TCP/IP is not supported by IPv6.

Network authentication

Using insecure network authentication methods may allow an adversary to gain unauthorised access to network traffic and services. To reduce this risk, only secure network authentication methods, ideally Kerberos, should be used for network authentication.

The following Group Policy settings can be implemented to configure Kerberos, and if required for legacy purposes, the use of NTLMv2.

NoLMHash policy

When Microsoft Windows hashes a password that is less than 15 characters, it stores both a LAN Manager hash (LM hash) and Windows NT hash (NT hash) in the local SAM database for local accounts, or in Activity Directory for domain accounts. The LM hash is significantly weaker than the NT hash and can easily be brute forced. To reduce this risk, the NoLMHash Policy should be implemented on all workstations and domain controllers. As the LM hash is designed for authentication of legacy Microsoft Windows operating systems, such as those prior to Microsoft Windows 2000, there shouldn’t be a business requirement for its use except in very rare circumstances.

The following Group Policy setting can be implemented to prevent the storage of LM hashes for passwords. All users should be encouraged to change their password once this Group Policy setting has been set as until they do they will remain vulnerable.

Operating system functionality

Leaving unneeded functionality in Microsoft Windows enabled can provide greater opportunities for potentially vulnerable or misconfigured functionality to be exploited by an adversary. To reduce this risk, unneeded functionality in Microsoft Windows should be disabled or removed.

Password and logon authentication policy

The use of weak passwords, such as eight character passwords with no complexity, can allow them to be brute forced within minutes using applications freely available on the web. To reduce this risk, a secure password policy should be implemented.

The following Group Policy settings can be implemented to achieve a secure single-factor password policy. Note, Group Policy settings for passwords used as part of multi-factor authentication may not need to be as stringent (e.g. a length of 6 characters without complexity).

Power management

One method of reducing power usage by workstations is to enter a sleep, hibernation or hybrid sleep state after a pre-defined period of inactivity. When a workstation enters a sleep state it maintains the contents of memory while powering down the rest of the workstation; with hibernation or hybrid sleep, it writes the contents of memory to the hard drive in a hibernation file (hiberfil.sys) and powers down the rest of the workstation. When this occurs, sensitive information such as encryption keys could either be retained in memory or written to the hard drive in a hibernation file. An adversary with physical access to the workstation and either the memory or hard drive can recover the sensitive information using forensic techniques. To reduce this risk, sleep, hibernation and hybrid sleep states should be disabled.

The following Group Policy settings can be implemented to ensure that sleep, hibernation and hybrid sleep states are disabled.

PowerShell

Allowing any PowerShell script to execute exposes a workstation to the risk that a malicious script may be unwittingly executed by a user. To reduce this risk, users should not have the ability to execute PowerShell scripts; however, if using PowerShell scripts is an essential business requirement, only signed scripts should be allowed to execute. Ensuring that only signed scripts are allowed to execute can provide a level of assurance that a script is trusted and has been endorsed as having a legitimate business purpose.

For more information on how to effectively implement PowerShell see the Securing PowerShell in the Enterprise publication.

The following Group Policy settings can be implemented to control the use of PowerShell scripts.

Registry editing tools

One method for malicious code to maintain persistence (i.e. remain after a workstation is rebooted) is to use administrative privileges to modify the registry (as standard privileges only allow viewing of the registry). To reduce this risk, users should not have the ability to modify the registry using registry editing tools (i.e. regedit) or to make silent changes to the registry (i.e. using .reg files).

The following Group Policy setting can be implemented to prevent users from viewing or modifying the registry using registry editing tools.

Remote Assistance

While Remote Assistance can be a useful business tool to allow system administrators to remotely administer workstations, it can also pose a risk. When a user has a problem with their workstation they can generate a Remote Assistance invitation. This invitation authorises anyone that has access to it to remotely control the workstation that issued the invitation. Invitations can be sent by email, instant messaging or saved to a file. If an adversary manages to intercept an invitation they will be able to use it to access the user’s workstation. Additionally, if network traffic on port 3389 is not blocked from reaching the internet, users may send Remote Assistance invitations over the internet which could allow for remote access to their workstation by an adversary. While Remote Assistance only grants access to the privileges of the user that generated the request, an adversary could install a key logging application on the workstation in preparation of a system administer using their privileged credentials to fix any problems. To reduce this risk, Remote Assistance should be disabled.

The following Group Policy settings can be implemented to disable Remote Assistance.

Remote Desktop Services

While remote desktop access may be convenient for legitimate users to access workstations across a network, it also allows an adversary to access other workstations once they have compromised an initial workstation and user’s credentials. This risk can be compounded if an adversary can compromise domain administrator credentials or common local administrator credentials. To reduce this risk, Remote Desktop Services should be disabled.

The following Group Policy settings can be implemented to disable Remote Desktop Services.

Alternatively, if it is an essential business requirement to use Remote Desktop Services, it should be configured in a manner that is as secure as possible and only on workstations and for users for which it is explicitly required.

The following Group Policy settings can be implemented to use Remote Desktop Services in as secure a manner as possible.

Remote Procedure Call

Remote Procedure Call (RPC) is a technique used for facilitating client and server application communications using a common interface. RPC is designed to make client and server interaction easier and safer by using a common library to handle tasks such as security, synchronisation and data flows. If unauthenticated communications are allowed between client and server applications, it could result in accidental disclosure of sensitive information or the failure to take advantage of RPC security functionality. To reduce this risk, all RPC clients should authenticate to RPC servers.

The following Group Policy setting can be implemented to ensure RPC clients authenticate to RPC servers.

Reporting system information

Microsoft Windows contains a number of in-built functions to, often automatically and transparently, report system information to Microsoft. This includes system errors and crash information as well as inventories of applications, files, devices and drivers on the system. If captured by an adversary, this information could expose potentially sensitive information on workstations. This information could also subsequently be used by an adversary to tailor malicious code to target specific workstations or users. To reduce this risk, all in-built functions that report potentially sensitive system information should be directed to a corporate Windows Error Reporting server.

The following Group Policy settings can be implemented to prevent potentially sensitive system information being reported to Microsoft.

Safe Mode

An adversary with standard user credentials that can boot into Microsoft Windows using Safe Mode, Safe Mode with Networking or Safe Mode with Command Prompt options may be able to bypass system protections and security functionality. To reduce this risk, users with standard credentials should be prevented from using Safe Mode options to log in.

The following registry entry can be implemented using Group Policy preferences to prevent non-administrators from using Safe Mode options.

Secure channel communications

Periodically, workstations connected to a domain will communicate with the domain controllers. If an adversary has access to unprotected network communications they may be able to capture or modify sensitive information communicated between workstations and the domain controllers. To reduce this risk, all secure channel communications should be signed and encrypted with strong session keys.

The following Group Policy settings can be implemented to ensure secure channel communications are appropriately signed and encrypted.

Security policies

By failing to comprehensively specify security policies, an adversary may be able to exploit weaknesses in a workstation’s Group Policy settings to gain access to sensitive information. To reduce this risk, security policies should be comprehensively specified.

The following Group Policy settings can be implemented, in addition to those specifically mentioned in other areas of this publication, to form a comprehensive set of security policies.

Server Message Block sessions

An adversary that has access to network communications may attempt to use session hijacking tools to interrupt, terminate or steal a Server Message Block (SMB) session. This could potentially allow an adversary to modify packets and forward them to a SMB server to perform undesirable actions or to pose as the server or client after a legitimate authentication has taken place to gain access to sensitive information. To reduce this risk, all communications between SMB clients and servers should be signed, with any passwords used appropriately encrypted.

The following Group Policy settings can be implemented to ensure communications between SMB clients and servers are secure. Note, the MS Security Guide Group Policy settings are available as part of the Microsoft Security Compliance Toolkit.

Session locking

An adversary with physical access to an unattended workstation with an unlocked session may attempt to inappropriately access sensitive information or conduct actions that won’t be attributed to them. To reduce this risk, a session lock should be configured to activate after a maximum of 15 minutes of user inactivity. Furthermore, be aware that information or alerts may be displayed on the lock screen. To reduce the risk of unauthorised information disclosure, minimise the amount of information that the lock screen is permitted to display.

The following Group Policy settings can be implemented to set session locks.

Software-based firewalls

Network firewalls often fail to prevent the propagation of malicious code on a network, or an adversary from extracting sensitive information, as they generally only control which ports or protocols can be used between segments on a network. Many forms of malicious code are designed specifically to take advantage of this by using common protocols such as HTTP, HTTPS, SMTP and DNS. To reduce this risk, software-based firewalls that filter both incoming and outgoing traffic should be appropriately implemented. Software-based firewalls are more effective than network firewalls as they can control which applications and services can communicate to and from workstations. The in-built Windows firewall can be used to control both inbound and outbound traffic for specific applications.

Sound Recorder

Sound Recorder is a feature of Microsoft Windows that allows audio from a device with a microphone to be recorded and saved as an audio file on the local hard drive. An adversary with remote access to a workstation can use this functionality to record sensitive conversations in the vicinity of the workstation. To reduce this risk, Sound Recorder should be disabled.

The following Group Policy setting can be implemented to disable the use of Sound Recorder.

Standard Operating Environment

When users are left to setup, configure and maintain their own workstations it can very easily lead to an inconsistent and insecure environment where particular workstations are more vulnerable than others. This inconsistent and insecure environment can easily allow an adversary to gain an initial foothold on a network. To reduce this risk, workstations should connect to a domain using a Standard Operating Environment that is centrally controlled and configured by experienced information technology and information security professionals. However, in some cases, cloud-based domain services may be more effective in deploying workstation configurations to a mobile and disparate workforce. In particular, security objectives may be achieved without the need to create ‘gold’ images and can offer more flexible enrolment processes. However, enrolment methods, such as Microsoft Intune self-enrolment, may introduce their own security risks, such as leaving behind local administrator accounts.

System backup and restore

An adversary that compromises a user account with privileges to backup files and directories can use this privilege to backup the contents of a workstation. This content can then be transferred to a non-domain connected workstation where the adversary has administrative access. From here an adversary can restore the contents and take ownership, thereby circumventing all original access controls that were in place. In addition, if a user has privileges to restore files and directories, an adversary could exploit this privilege by using it to either restore previous versions of files that may have been removed by system administrators as part of malicious code removal activities or to replace existing files with malicious variants. To reduce this risk, the ability to use backup and restore functionality should be limited to administrators.

The following Group Policy settings can be implemented to control the use of backup and restore functionality.

System cryptography

By default, when cryptographic keys are stored in Microsoft Windows, users can access them without first entering a password to unlock the certificate store. An adversary that compromises a workstation, or gains physical access to an unlocked workstation, can use these user keys to access sensitive information or resources that are cryptographically protected. To reduce this risk, strong encryption algorithms and strong key protection should be used on workstations.

The following Group Policy settings can be implemented to ensure strong encryption algorithms and strong key protection is used.

User rights policies

By failing to comprehensively specify user rights policies, an adversary may be able to exploit weaknesses in a workstation’s Group Policy settings to gain access to sensitive information. To reduce this risk, user rights policies should be comprehensively specified.

The following Group Policy settings can be implemented, in addition to those specifically mentioned in other areas of this publication, to form a comprehensive set of user rights policies.

Virtualised web and email access

An adversary can often deliver malicious code directly to workstations via external web and email access. Once a workstation has been exploited, an adversary can use these same communication paths for bi-directional communications to control their malicious code. To reduce this risk, web and email access on workstations should occur through a non-persistent virtual environment (i.e. using virtual desktops or virtual applications). When using a virtual environment, workstations will receive additional protection against intrusion attempts targeted at exploiting security vulnerabilities in web browsers and email clients as any attempts, if successful, will execute in a non-persistent virtual environment rather than on a local workstation.

Web Proxy Auto Discovery protocol

The Web Proxy Auto Discovery (WPAD) protocol assists with the automatic detection of proxy settings for web browsers. Unfortunately, WPAD has suffered from a number of severe security vulnerabilities. Organisations that do not rely on the use of the WPAD protocol should disable it. This can be achieved by modifying each workstation’s host file at %SystemDrive%\Windows\System32\Drivers\etc\hosts to create the following entry: 255.255.255.255 wpad.

Windows Remote Management

Windows Remote Management (WinRM) is the Microsoft implementation of the WS-Management Protocol which was developed as a public standard for remotely exchanging management data between devices that implement the protocol. If appropriate authentication and encryption is not implemented for this protocol, traffic may be subject to inception by an adversary. To reduce this risk, Windows Remote Management should be securely configured.

The following Group Policy settings can be implemented to secure the use of the Windows Remote Management.

Windows Remote Shell access

When Windows Remote Shell is enabled it can allow an adversary to remotely execute scripts and commands on workstations. To reduce this risk, Windows Remote Shell should be disabled.

The following Group Policy setting can be implemented to disable Windows Remote Shell access.

Windows Search and Cortana

As part of the in-built search functionality of Microsoft Windows, users can search for web results in addition to local workstation results. This functionality if used could result in the accidental disclosure of sensitive information if sensitive terms are searched for automatically on the web in addition to the local workstation. To reduce this risk, the ability to automatically search the web should be disabled.

The following Group Policy settings can be implemented to prevent web search results being returned for any user search terms.

Low priorities

The following recommendations, listed in alphabetical order, should be treated as low priorities when hardening Microsoft Windows 10 workstations.

Displaying file extensions

When extensions for known file types are hidden, an adversary can more easily use social engineering techniques to convince users to execute malicious email attachments. For example, a file named vulnerability_assessment.pdf.exe could appear as vulnerability_assessment.pdf to a user. To reduce this risk, hiding extensions for known file types should be disabled. Showing extensions for all known file types, in combination with user education and awareness of dangerous email attachment file types, can help reduce the risk of users executing malicious email attachments.

The following registry entry can be implemented using Group Policy preferences to prevent extensions for known file types from being hidden.

File and folder security properties

By default, all users have the ability to view security properties of files and folders. This includes the security properties associated with files and folders as well as users and groups that they relate to. An adversary could use this information to target specific accounts that have access to sensitive information. To reduce this risk, users should not have the ability to view security properties of files and folders.

The following Group Policy setting can be implemented to disable users’ access to the security tab in file and folder properties in File Explorer.

Location awareness

When users interact with the internet their workstations often automatically provide geo-location details to websites or online services to assist them in tailoring content specific to the user’s geographical region (i.e. the city they are accessing the internet from). This information can be captured by an adversary to determine the location of a specific user. To reduce this risk, location services in the operating system and applications should be disabled.

The following Group Policy settings can be implemented to disable location services within the operating system.

Microsoft Store

Whilst applications in the Microsoft Store are vetted by Microsoft, there is still a risk that users given access to the Microsoft Store could download and install potentially malicious applications or applications that cause conflicts with other endorsed applications on their workstation. To reduce this risk, access to the Microsoft Store should be disabled.

The following Group Policy settings can be implemented to prevent Microsoft Store access.

Resultant Set of Policy reporting

By default, all users have the ability to generate Resultant Set of Policy (RSOP) reports which allows them to view the Group Policy settings being applied to their workstation and user account. This information could be used by an adversary to determine misconfigurations or weaknesses in Group Policy settings being applied to the workstation or the user account. To reduce this risk, users should not have the ability to generate RSOP reports.

The following Group Policy setting can be implemented to disable users’ ability to generate RSOP reports.

Further information

The Information Security Manual is a cyber security framework that organisations can apply to protect their systems and data from cyber threats. The advice in the Strategies to Mitigate Cyber Security Incidents, along with its Essential Eight, complements this framework.

Contact details

If you have any questions regarding this guidance you can write to us or call us on 1300 CYBER1 (1300 292 371).