How the attack works
The security investigation tool of Google Workspace lets administrators search log and message data across the whole organization and act on the results. Google lists access to data about Gmail messages, including email content, among its uses. A search returns message headers. To open a message, the administrator needs the View sensitive content privilege, and a super admin must have turned on the View sensitive content setting of the tool, which covers email, chat messages and chat attachments. The same settings window can require a written justification before content is shown.
The tool is built for responders, and it gives the same reach to whoever holds a responder's account. An attacker with the security center privileges searches the Gmail message data of the organization, opens the messages that matter and downloads their attachments. No mailbox is signed in to, no delegation is added and no export job is created, so the usual signs of mail collection are missing. If the setting is off, a super admin account can turn it on first.
Google states that administrator queries and actions in the investigation tool are recorded in the Admin log events, that these logs are always written and cannot be turned off, and that the reason entered for viewing a message is recorded there. Its list of Cloud Audit Logs methods for the Admin audit holds securityInvestigationContentAccess, securityInvestigationDownloadAttachment and securityInvestigationSettingUpdate. Google's event reference has no entry for them, so the rules on this page match on the method names alone.
What an attacker needs
- An admin account with the Security center privileges for audit and investigation, including View sensitive content, or a super admin account.
- The View sensitive content setting of the investigation tool turned on, which only a super admin can do.
- An edition that includes the tool and the viewing of message content. Google lists the supported editions on its help pages.
Impact
The attacker reads mail of users across the organization, chosen by search, and can download attachments. After opening a message the tool also offers to delete it or move it, which can remove evidence from a mailbox. The mailbox owners see nothing.
Detection rules
Content Accessed or Attachment Downloaded With the Security Investigation Tool in Google Workspace
Detects an administrator accessing content or downloading an attachment through the security investigation tool of a Google Workspace account (Cloud Audit Logs methods securityInvestigationContentAccess and securityInvestigationDownloadAttachment of admin.googleapis.com). Google documents that the tool can show the content of Gmail messages, chat messages and attachments to administrators who hold the View sensitive content privilege, and that the reason entered for viewing a message is recorded in the Admin log events. The rule matches on the method names ONLY. Google lists the two methods on its Cloud Audit Logs page but its event reference has no entry for them, so their parameters are not documented and the rule cannot tell which message or data source was opened. The level is one step lower than the access deserves for that reason. Viewing content is a deliberate, rare action even for a security team, so match every event with an incident ticket. The admin is in principalEmail. The entries reach Cloud Audit Logs as Admin Activity entries at organization level once Google Workspace data sharing with Google Cloud is enabled, and reach the SIEM only when the organization-level logs are routed to it.
title: Content Accessed or Attachment Downloaded With the Security Investigation Tool in Google Workspace
id: 3d140437-0c21-464c-b5e4-5feaa4a0be0c
status: experimental
description: >-
Detects an administrator accessing content or downloading an attachment
through the security investigation tool of a Google Workspace account
(Cloud Audit Logs methods securityInvestigationContentAccess and
securityInvestigationDownloadAttachment of admin.googleapis.com). Google
documents that the tool can show the content of Gmail messages, chat
messages and attachments to administrators who hold the View sensitive
content privilege, and that the reason entered for viewing a message is
recorded in the Admin log events. The rule matches on the method names
ONLY. Google lists the two methods on its Cloud Audit Logs page but its
event reference has no entry for them, so their parameters are not
documented and the rule cannot tell which message or data source was
opened. The level is one step lower than the access deserves for that
reason. Viewing content is a deliberate, rare action even for a security
team, so match every event with an incident ticket. The admin is in
principalEmail. The entries reach Cloud Audit Logs as Admin Activity
entries at organization level once Google Workspace data sharing with
Google Cloud is enabled, and reach the SIEM only when the
organization-level logs are routed to it.
references:
- https://docs.cloud.google.com/logging/docs/audit/gsuite-audit-logging
- https://knowledge.workspace.google.com/admin/security/use-the-investigation-tool-to-view-sensitive-content
- https://knowledge.workspace.google.com/admin/security/admin-auditing-for-the-security-center
- https://knowledge.workspace.google.com/admin/security/admin-privileges-for-the-security-investigation-tool
- https://attack.mitre.org/techniques/T1114/002/
author: Forestall
date: 2026-11-18
tags:
- attack.collection
- attack.t1114.002
logsource:
product: gcp
service: gcp.audit
detection:
selection:
protoPayload.serviceName: admin.googleapis.com
protoPayload.methodName:
- google.admin.AdminService.securityInvestigationContentAccess
- google.admin.AdminService.securityInvestigationDownloadAttachment
condition: selection
falsepositives:
- Incident responders who open a reported phishing message or the content behind a data protection rule. Each access should have a ticket and a justification text. Keep a list of the administrators who are allowed to do this and review every event from anyone else first.
level: medium
False positives
- Incident responders who open a reported phishing message or the content behind a data protection rule. Each access should have a ticket and a justification text. Keep a list of the administrators who are allowed to do this and review every event from anyone else first.
Security Investigation Tool Settings Updated in Google Workspace
Detects a change of the settings of the security investigation tool of a Google Workspace account (Cloud Audit Logs method securityInvestigationSettingUpdate of admin.googleapis.com). Google documents the settings a super admin controls there. They are whether administrators may view sensitive content such as email and chat messages, whether a justification must be entered before content is shown or before an action, and whether bulk actions need a reviewer. Turning content viewing on, or the justification and reviewer requirements off, prepares the tool for reading mail unnoticed. The rule matches on the method name ONLY. Google lists the method on its Cloud Audit Logs page but its event reference has no entry for it, so the rule cannot tell which setting changed or in which direction, and the level is low for that reason. Read the entry in the Admin log events of the Admin console. The settings change rarely. The admin is in principalEmail. The entry reaches Cloud Audit Logs as an Admin Activity entry at organization level once Google Workspace data sharing with Google Cloud is enabled, and reaches the SIEM only when the organization-level logs are routed to it.
title: Security Investigation Tool Settings Updated in Google Workspace
id: 005dc57d-7e78-4aeb-8274-c84dd1111e6f
status: experimental
description: >-
Detects a change of the settings of the security investigation tool of a
Google Workspace account (Cloud Audit Logs method
securityInvestigationSettingUpdate of admin.googleapis.com). Google
documents the settings a super admin controls there. They are whether
administrators may view sensitive content such as email and chat
messages, whether a justification must be entered before content is
shown or before an action, and whether bulk actions need a reviewer.
Turning content viewing on, or the justification and reviewer
requirements off, prepares the tool for reading mail unnoticed. The rule
matches on the method name ONLY. Google lists the method on its Cloud
Audit Logs page but its event reference has no entry for it, so the rule
cannot tell which setting changed or in which direction, and the level is
low for that reason. Read the entry in the Admin log events of the Admin
console. The settings change rarely. The admin is in principalEmail. The
entry reaches Cloud Audit Logs as an Admin Activity entry at organization
level once Google Workspace data sharing with Google Cloud is enabled,
and reaches the SIEM only when the organization-level logs are routed to
it.
references:
- https://docs.cloud.google.com/logging/docs/audit/gsuite-audit-logging
- https://knowledge.workspace.google.com/admin/security/configure-settings-for-your-investigations
- https://knowledge.workspace.google.com/admin/security/admin-auditing-for-the-security-center
- https://attack.mitre.org/techniques/T1114/002/
author: Forestall
date: 2026-11-18
tags:
- attack.collection
- attack.t1114.002
logsource:
product: gcp
service: gcp.audit
detection:
selection:
protoPayload.serviceName: admin.googleapis.com
protoPayload.methodName: google.admin.AdminService.securityInvestigationSettingUpdate
condition: selection
falsepositives:
- A super admin who sets the tool up, or who tightens it by requiring a justification or a reviewer, which matches as well. Verify every event with the super admin who made it and compare the settings window with your record.
level: low
False positives
- A super admin who sets the tool up, or who tightens it by requiring a justification or a reviewer, which matches as well. Verify every event with the super admin who made it and compare the settings window with your record.
Detection logic
Patterns that need more than one event, a baseline or knowledge of your own environment. Each is a hypothesis to build and tune in your SIEM, not a ready rule.
Investigation settings updated and message content accessed by the same admin soon after
A responder works with settings that were decided long ago. An account that changes the settings of the investigation tool and then opens message content has most likely just given itself the access, which fits a compromised super admin more than an investigation.
- The settings of the tool change:
google.admin.AdminService.securityInvestigationSettingUpdateofadmin.googleapis.com. Keep the admin inauthenticationInfo.principalEmail. - Within 24 hours the same
principalEmailwritesgoogle.admin.AdminService.securityInvestigationContentAccessorsecurityInvestigationDownloadAttachment. - Raise the pair and read both entries in the Admin log events of the Admin console for the setting that changed and the justification that was entered.
Eventsgoogle.admin.AdminService.securityInvestigationSettingUpdategoogle.admin.AdminService.securityInvestigationContentAccessgoogle.admin.AdminService.securityInvestigationDownloadAttachment
LimitsGoogle lists these methods without describing their parameters, so the steps use the method name and the caller only. A security lead who turns the setting on for a real incident and reads the first message matches as well. An attacker who uses a second account for the setting is not found. The 24 hours are a starting point.
Triage
- Identify the admin (principalEmail) and the time, and ask for the incident or legal request that required reading the message.
- Open the same entry in the Admin log events of the Admin console and read the justification the admin entered and the details Google shows for the event.
- Review the investigation tool queries of that admin before the access (google.admin.AdminService.securityInvestigationQuery) to see how often they searched and when.
- Count the content accesses and attachment downloads of the admin in the period. A responder opens a few messages of one campaign. Many messages of different senders and topics point to collection.
- Check whether the investigation settings or the admin's role changed shortly before, and the recent sign-ins of the admin account.
Response
Once the attack is confirmed:
- Remove the security center privileges from the account, or suspend it, and reset its password and sign-in cookies.
- Turn the View sensitive content setting off, or restore the justification requirement, if it was changed for this access.
- Establish which messages and attachments were opened and tell the owners of the mailboxes and the data protection officer as your process requires.
- Check the affected mailboxes for messages that were deleted or moved through the tool after they were read.
- Review the admin roles that include the Security center privileges and remove View sensitive content where it is not needed.
Mitigation
- Keep the View sensitive content setting off unless the security team needs it, and turn on the requirement to enter a justification before content is shown.
- Give the View sensitive content privilege to a small named group and keep it out of general admin roles. Super admins hold the audit and investigation privileges by default, so keep super admins few.
- Turn on the Require reviewer setting for bulk actions in the tool.
- Review the Admin log events of the investigation tool regularly, with the justification texts, against your incident tickets.
- Share Workspace data with Google Cloud and route the organization-level audit logs to the SIEM.
Frequently asked questions
Who can read other people's mail with the security investigation tool?
Google ties it to two things. The administrator needs the View sensitive content privilege under the Security Center audit and investigation privileges, and a super admin has to turn on the View sensitive content setting in the settings of the tool. Google states that super admins have the audit and investigation privileges by default. The viewing of message content is also limited to certain editions.
Does the audit entry show which message was read and why?
Google states that the reason an admin enters before viewing a message is recorded in the Admin log events. Its event reference has no entry for the investigation tool events, so the parameters of the Cloud Audit Logs entry are not documented. Read an entry from your own tenant to see which details it carries before you build on them.
Can an administrator turn this logging off?
Google states that the audit logs for security center tasks are always written and that the events cannot be configured, excluded or disabled. What can be lost is the copy outside Google. The entries reach Cloud Audit Logs only while Workspace data sharing with Google Cloud is on, and the SIEM only while the organization-level logs are routed to it.
How does this differ from a mailbox export by an admin?
An export copies a whole mailbox and leaves an export request. The investigation tool reads single messages in place after a search of the organization's message data, so it suits an attacker who knows what they are looking for. The two leave different audit events, and the library has a separate page for mailbox exports and data transfers.
References
- MITRE ATT&CK T1114.002: Email Collection: Remote Email Collection
- About the security investigation tool
- Use the investigation tool to view sensitive content
- Configure settings for your investigations
- Admin privileges for the security investigation tool
- Admin auditing for the security center
- Audit logs for Google Workspace (Cloud Audit Logs, method list)