How the attack works
Google analyses sign-ins to Workspace and Cloud Identity accounts, web sign-ins and programmatic ones alike. A sign-in attempt that does not match the user's normal behavior, for example one from an unusual location, counts as suspicious. In most cases the user first sees a login challenge, and Google raises the warning when the challenge fails or is abandoned. Google also acts on what it learns outside the sign-in itself. When it detects compromised credentials it requires a password reset before the user can sign in again, and when activity suggests that an account has been taken over it suspends the account.
An attacker who signs in with a stolen, guessed or reused password produces exactly these events. Each one means that somebody other than the owner probably has the password, or tried to use the account, and the same identities sign in to Google Cloud. The Login audit log records them as account warning events, namely suspicious_login, suspicious_login_less_secure_app, suspicious_programmatic_login, account_disabled_password_leak and account_disabled_hijacked, and as the attack warning event gov_attack_warning. The same log has Google's other suspensions, for spam sent from the account (account_disabled_spamming, account_disabled_spamming_through_relay) and without a stated reason (account_disabled_generic), and the event risky_sensitive_action_blocked, written when Google refuses a sensitive action such as disabling 2-Step Verification because the user did not pass its challenge.
Organizations that share Workspace data with Google Cloud receive them in Cloud Audit Logs at organization level as Data Access entries of login.googleapis.com, with method names such as google.login.LoginService.suspiciousLogin and google.login.LoginService.accountDisabledHijacked.
What an attacker needs
- A user's password, obtained through phishing, malware or reuse of the password on a site that was breached.
- An account without 2-Step Verification, where the password alone completes the sign-in.
Impact
If the sign-in went through before Google acted, the attacker had the user's mail, files and directory, and every admin role in Workspace and IAM role in Google Cloud that the account holds. Where Google blocked the attempt, the password is still known to the attacker and can be tried again from another place or against other services where the user reused it.
Detection rules
Account Disabled by Google for a Leaked Password in Google Workspace
Detects Google disabling a Google Workspace or Cloud Identity account because it has become aware that someone else knows the password (Login audit event account_disabled_password_leak). Google requires a password reset before the user can sign in again. The affected account is in the event parameters (affected_email_address in protoPayload.metadata). Check whether the password was used before Google acted. The entry reaches Cloud Audit Logs as a Data Access entry of login.googleapis.com at organization level only when Google Workspace data sharing with Google Cloud is enabled, and reaches the SIEM only when the organization-level Data Access logs are routed to it.
title: Account Disabled by Google for a Leaked Password in Google Workspace
id: de303eb2-2308-41b0-8c0a-8094cbaf69f4
status: experimental
description: >-
Detects Google disabling a Google Workspace or Cloud Identity account
because it has become aware that someone else knows the password (Login
audit event account_disabled_password_leak). Google requires a password
reset before the user can sign in again. The affected account is in the
event parameters (affected_email_address in protoPayload.metadata). Check
whether the password was used before Google acted. The entry reaches Cloud
Audit Logs as a Data Access entry of login.googleapis.com at organization
level only when Google Workspace data sharing with Google Cloud is enabled,
and reaches the SIEM only when the organization-level Data Access logs are
routed to it.
references:
- https://docs.cloud.google.com/logging/docs/audit/gsuite-audit-logging
- https://developers.google.com/workspace/admin/reports/v1/appendix/activity/login
- https://knowledge.workspace.google.com/admin/security/view-alert-details
- https://attack.mitre.org/techniques/T1078/004/
author: Forestall
date: 2026-11-16
tags:
- attack.initial-access
- attack.t1078.004
logsource:
product: gcp
service: gcp.audit
detection:
selection:
protoPayload.serviceName: login.googleapis.com
protoPayload.methodName: google.login.LoginService.accountDisabledPasswordLeak
condition: selection
falsepositives:
- None expected as a false signal, since the event is Google's own finding. A password that leaked from another site where the user reused it produces the event without any sign-in to the organization, so the urgency depends on what the account's recent sign-ins show.
level: medium
False positives
- None expected as a false signal, since the event is Google's own finding. A password that leaked from another site where the user reused it produces the event without any sign-in to the organization, so the urgency depends on what the account's recent sign-ins show.
Account Suspended by Google for Spamming in Google Workspace
Detects Google disabling a Google Workspace or Cloud Identity account because it has become aware that the account was used to send spam, either directly (Login audit event account_disabled_spamming) or through the SMTP relay service (account_disabled_spamming_through_relay). A mailbox that starts sending spam is often one whose password someone else has, so check the account for a takeover before treating the event as a mail problem. The affected account is in the event parameters (affected_email_address in protoPayload.metadata). The entry reaches Cloud Audit Logs as a Data Access entry of login.googleapis.com at organization level only when Google Workspace data sharing with Google Cloud is enabled, and reaches the SIEM only when the organization-level Data Access logs are routed to it.
title: Account Suspended by Google for Spamming in Google Workspace
id: 00272f89-ee1f-4625-a38c-7cfd352d7591
status: experimental
description: >-
Detects Google disabling a Google Workspace or Cloud Identity account
because it has become aware that the account was used to send spam, either
directly (Login audit event account_disabled_spamming) or through the SMTP
relay service (account_disabled_spamming_through_relay). A mailbox that
starts sending spam is often one whose password someone else has, so check
the account for a takeover before treating the event as a mail problem.
The affected account is in the event parameters
(affected_email_address in protoPayload.metadata). The entry reaches Cloud
Audit Logs as a Data Access entry of login.googleapis.com at organization
level only when Google Workspace data sharing with Google Cloud is enabled,
and reaches the SIEM only when the organization-level Data Access logs are
routed to it.
references:
- https://docs.cloud.google.com/logging/docs/audit/gsuite-audit-logging
- https://developers.google.com/workspace/admin/reports/v1/appendix/activity/login
- https://developers.google.com/workspace/admin/alertcenter/reference/alert-types
- https://attack.mitre.org/techniques/T1078/004/
author: Forestall
date: 2026-11-18
tags:
- attack.initial-access
- attack.t1078.004
logsource:
product: gcp
service: gcp.audit
detection:
selection:
protoPayload.serviceName: login.googleapis.com
protoPayload.methodName:
- google.login.LoginService.accountDisabledSpamming
- google.login.LoginService.accountDisabledSpammingThroughRelay
condition: selection
falsepositives:
- A user or an application that sent bulk mail from the account on purpose, such as a newsletter or a misconfigured device that relays through Google. Check what was sent and from where before restoring the account.
level: medium
False positives
- A user or an application that sent bulk mail from the account on purpose, such as a newsletter or a misconfigured device that relays through Google. Check what was sent and from where before restoring the account.
Account Suspended by Google for Suspected Compromise in Google Workspace
Detects Google suspending a Google Workspace or Cloud Identity account because it detected suspicious activity indicating that the account might have been compromised (Login audit event account_disabled_hijacked). Treat it as a likely takeover: review what the account did before the suspension and secure it before restoring it. The affected account is in the event parameters (affected_email_address in protoPayload.metadata), with the time of the sign-in in login_timestamp. The entry reaches Cloud Audit Logs as a Data Access entry of login.googleapis.com at organization level only when Google Workspace data sharing with Google Cloud is enabled, and reaches the SIEM only when the organization-level Data Access logs are routed to it.
title: Account Suspended by Google for Suspected Compromise in Google Workspace
id: d9408eed-0cbc-4d1b-ab2b-04e38565a6b2
status: experimental
description: >-
Detects Google suspending a Google Workspace or Cloud Identity account
because it detected suspicious activity indicating that the account might
have been compromised (Login audit event account_disabled_hijacked). Treat
it as a likely takeover: review what the account did before the suspension
and secure it before restoring it. The affected account is in the event
parameters (affected_email_address in protoPayload.metadata), with the time
of the sign-in in login_timestamp. The entry reaches Cloud Audit Logs as a
Data Access entry of login.googleapis.com at organization level only when
Google Workspace data sharing with Google Cloud is enabled, and reaches the
SIEM only when the organization-level Data Access logs are routed to it.
references:
- https://docs.cloud.google.com/logging/docs/audit/gsuite-audit-logging
- https://developers.google.com/workspace/admin/reports/v1/appendix/activity/login
- https://knowledge.workspace.google.com/admin/security/view-alert-details
- https://attack.mitre.org/techniques/T1078/004/
author: Forestall
date: 2026-11-16
tags:
- attack.initial-access
- attack.t1078.004
logsource:
product: gcp
service: gcp.audit
detection:
selection:
protoPayload.serviceName: login.googleapis.com
protoPayload.methodName: google.login.LoginService.accountDisabledHijacked
condition: selection
falsepositives:
- Unusual but legitimate activity by the account owner that Google's analysis read as a takeover. Confirm with the user through a second channel before restoring the account. Suspensions for spam have their own events and do not match this rule.
level: high
False positives
- Unusual but legitimate activity by the account owner that Google's analysis read as a takeover. Confirm with the user through a second channel before restoring the account. Suspensions for spam have their own events and do not match this rule.
Account Suspended by Google Without a Stated Reason in Google Workspace
Detects the Login audit event account_disabled_generic, which Google's reference titles "User suspended" and renders only as the account being disabled. Unlike the events for a leaked password, a suspected compromise or spam, it gives no reason, so it has to be read together with the account's other login entries. The affected account is in the event parameters (affected_email_address in protoPayload.metadata). A suspension by an administrator is a different event, written to the Admin audit log, and does not match. The entry reaches Cloud Audit Logs as a Data Access entry of login.googleapis.com at organization level only when Google Workspace data sharing with Google Cloud is enabled, and reaches the SIEM only when the organization-level Data Access logs are routed to it.
title: Account Suspended by Google Without a Stated Reason in Google Workspace
id: 266fd62d-d6dc-4237-a433-baa85f6bdad7
status: experimental
description: >-
Detects the Login audit event account_disabled_generic, which Google's
reference titles "User suspended" and renders only as the account being
disabled. Unlike the events for a leaked password, a suspected compromise
or spam, it gives no reason, so it has to be read together with the
account's other login entries. The affected account is in the event
parameters (affected_email_address in protoPayload.metadata). A suspension
by an administrator is a different event, written to the Admin audit log,
and does not match. The entry reaches Cloud Audit Logs as a Data Access
entry of login.googleapis.com at organization level only when Google
Workspace data sharing with Google Cloud is enabled, and reaches the SIEM
only when the organization-level Data Access logs are routed to it.
references:
- https://docs.cloud.google.com/logging/docs/audit/gsuite-audit-logging
- https://developers.google.com/workspace/admin/reports/v1/appendix/activity/login
- https://developers.google.com/workspace/admin/alertcenter/reference/alert-types
- https://attack.mitre.org/techniques/T1078/004/
author: Forestall
date: 2026-11-18
tags:
- attack.initial-access
- attack.t1078.004
logsource:
product: gcp
service: gcp.audit
detection:
selection:
protoPayload.serviceName: login.googleapis.com
protoPayload.methodName: google.login.LoginService.accountDisabledGeneric
condition: selection
falsepositives:
- Google does not state the reason in this event, so it can stand for causes that have nothing to do with a stolen password. Read the alert in the Alert Center and the account's recent sign-ins before deciding how urgent it is.
level: low
False positives
- Google does not state the reason in this event, so it can stand for causes that have nothing to do with a stolen password. Read the alert in the Alert Center and the account's recent sign-ins before deciding how urgent it is.
Government-Backed Attack Warning for a Google Workspace User
Detects Google warning that a user of a Google Workspace or Cloud Identity account might have been targeted by a government-backed attack (Login audit event gov_attack_warning). The warning says the user is being targeted, for example by an attempt to steal the password, not that the attempt succeeded. The user is the actor of the event (principalEmail). The entry reaches Cloud Audit Logs as a Data Access entry of login.googleapis.com at organization level only when Google Workspace data sharing with Google Cloud is enabled, and reaches the SIEM only when the organization-level Data Access logs are routed to it.
title: Government-Backed Attack Warning for a Google Workspace User
id: 1db3df12-c0c9-4c04-8409-46472139cb5e
status: experimental
description: >-
Detects Google warning that a user of a Google Workspace or Cloud Identity
account might have been targeted by a government-backed attack (Login audit
event gov_attack_warning). The warning says the user is being targeted, for
example by an attempt to steal the password, not that the attempt
succeeded. The user is the actor of the event (principalEmail). The entry
reaches Cloud Audit Logs as a Data Access entry of login.googleapis.com at
organization level only when Google Workspace data sharing with Google
Cloud is enabled, and reaches the SIEM only when the organization-level
Data Access logs are routed to it.
references:
- https://docs.cloud.google.com/logging/docs/audit/gsuite-audit-logging
- https://developers.google.com/workspace/admin/reports/v1/appendix/activity/login
- https://knowledge.workspace.google.com/admin/security/view-alert-details
- https://attack.mitre.org/techniques/T1078/004/
author: Forestall
date: 2026-11-16
tags:
- attack.initial-access
- attack.t1078.004
logsource:
product: gcp
service: gcp.audit
detection:
selection:
protoPayload.serviceName: login.googleapis.com
protoPayload.methodName: google.login.LoginService.govAttackWarning
condition: selection
falsepositives:
- None expected as a false signal, since the event is Google's own warning. It does not show that the account was accessed. Reset the user's password, check the account's 2-Step Verification and review its recent sign-ins.
level: high
False positives
- None expected as a false signal, since the event is Google's own warning. It does not show that the account was accessed. Reset the user's password, check the account's 2-Step Verification and review its recent sign-ins.
Sensitive Account Action Blocked by Google in Google Workspace
Detects Google refusing a sensitive action in a signed-in Google Workspace or Cloud Identity account (Login audit event risky_sensitive_action_blocked, rendered as the actor "wasn't allowed to attempt sensitive action"). Google challenges a user who attempts an action it considers sensitive, such as disabling 2-Step Verification, allowing an app to access Google data or changing the recovery email or phone, and disallows the action when the challenge is not passed. A session that is already signed in and then fails that check is what a stolen session or password looks like when the thief tries to secure their hold. The account is in principalEmail and the action in the event parameter sensitive_action_name in protoPayload.metadata. The event for an action that was allowed (risky_sensitive_action_allowed) does not match. The entry reaches Cloud Audit Logs as a Data Access entry of login.googleapis.com at organization level only when Google Workspace data sharing with Google Cloud is enabled, and reaches the SIEM only when the organization-level Data Access logs are routed to it.
title: Sensitive Account Action Blocked by Google in Google Workspace
id: 27ba9836-c412-40a8-804b-1edb44841aba
status: experimental
description: >-
Detects Google refusing a sensitive action in a signed-in Google Workspace
or Cloud Identity account (Login audit event
risky_sensitive_action_blocked, rendered as the actor "wasn't allowed to
attempt sensitive action"). Google challenges a user who attempts an action
it considers sensitive, such as disabling 2-Step Verification, allowing an
app to access Google data or changing the recovery email or phone, and
disallows the action when the challenge is not passed. A session that is
already signed in and then fails that check is what a stolen session or
password looks like when the thief tries to secure their hold. The account
is in principalEmail and the action in the event parameter
sensitive_action_name in protoPayload.metadata. The event for an action
that was allowed (risky_sensitive_action_allowed) does not match. The entry
reaches Cloud Audit Logs as a Data Access entry of login.googleapis.com at
organization level only when Google Workspace data sharing with Google
Cloud is enabled, and reaches the SIEM only when the organization-level
Data Access logs are routed to it.
references:
- https://docs.cloud.google.com/logging/docs/audit/gsuite-audit-logging
- https://developers.google.com/workspace/admin/reports/v1/appendix/activity/login
- https://knowledge.workspace.google.com/admin/security/protect-google-workspace-accounts-with-security-challenges
- https://attack.mitre.org/techniques/T1078/004/
author: Forestall
date: 2026-11-18
tags:
- attack.initial-access
- attack.t1078.004
logsource:
product: gcp
service: gcp.audit
detection:
selection:
protoPayload.serviceName: login.googleapis.com
protoPayload.methodName: google.login.LoginService.riskySensitiveActionBlocked
condition: selection
falsepositives:
- The account owner on a new device or network who cannot complete the challenge. Google's help page also describes users who cannot verify until a device, phone number or security key has been associated with the account for at least 7 days. Confirm with the user which action they attempted.
level: medium
False positives
- The account owner on a new device or network who cannot complete the challenge. Google's help page also describes users who cannot verify until a device, phone number or security key has been associated with the account for at least 7 days. Confirm with the user which action they attempted.
Suspicious Login Flagged by Google in Google Workspace
Detects the three Login audit events with which Google reports a suspicious sign-in to a Google Workspace or Cloud Identity account: suspicious_login, suspicious_login_less_secure_app and suspicious_programmatic_login. Google's reference titles them as blocked logins. The attempt did not match the user's normal behavior, so the password is probably known to someone else. The affected account is in the event parameters (affected_email_address in protoPayload.metadata), with the time of the flagged sign-in in login_timestamp. The entry reaches Cloud Audit Logs as a Data Access entry of login.googleapis.com at organization level only when Google Workspace data sharing with Google Cloud is enabled, and reaches the SIEM only when the organization-level Data Access logs are routed to it.
title: Suspicious Login Flagged by Google in Google Workspace
id: 647db8ef-ae0f-462c-8190-fb0caa7b5d4b
status: experimental
description: >-
Detects the three Login audit events with which Google reports a suspicious
sign-in to a Google Workspace or Cloud Identity account: suspicious_login,
suspicious_login_less_secure_app and suspicious_programmatic_login. Google's
reference titles them as blocked logins. The attempt did not match the
user's normal behavior, so the password is probably known to someone else.
The affected account is in the event parameters (affected_email_address in
protoPayload.metadata), with the time of the flagged sign-in in
login_timestamp. The entry reaches Cloud Audit Logs as a Data Access entry
of login.googleapis.com at organization level only when Google Workspace
data sharing with Google Cloud is enabled, and reaches the SIEM only when
the organization-level Data Access logs are routed to it.
references:
- https://docs.cloud.google.com/logging/docs/audit/gsuite-audit-logging
- https://developers.google.com/workspace/admin/reports/v1/appendix/activity/login
- https://knowledge.workspace.google.com/admin/security/view-alert-details
- https://attack.mitre.org/techniques/T1078/004/
author: Forestall
date: 2026-11-16
tags:
- attack.initial-access
- attack.t1078.004
logsource:
product: gcp
service: gcp.audit
detection:
selection:
protoPayload.serviceName: login.googleapis.com
protoPayload.methodName:
- google.login.LoginService.suspiciousLogin
- google.login.LoginService.suspiciousLoginLessSecureApp
- google.login.LoginService.suspiciousProgrammaticLogin
condition: selection
falsepositives:
- Users who travel, change devices or networks, or use an old mail client, and who fail or abandon the login challenge. Confirm the attempt with the user, and give priority to administrators and to accounts that hold IAM roles in Google Cloud.
level: medium
False positives
- Users who travel, change devices or networks, or use an old mail client, and who fail or abandon the login challenge. Confirm the attempt with the user, and give priority to administrators and to accounts that hold IAM roles in Google Cloud.
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.
Sign-in flagged by Google followed by a change that keeps access to the account
A flagged sign-in alone leaves open whether the attacker got in. What an intruder does next settles it. They disable 2-Step Verification, change the password or the recovery details, or forward the mail out of the domain. One of those changes on the same account shortly after Google flagged a sign-in is a takeover.
- Google flags a sign-in: one of the three suspicious login events of
google.login.LoginService, with the account in the event parameteraffected_email_address, or aloginSuccessentry whose parameteris_suspiciousis true, with the account inauthenticationInfo.principalEmail. - Within 24 hours the same account is the actor of
google.login.LoginService.2svDisable,passwordEdit,recoveryEmailEdit,recoveryPhoneEditoremailForwardingOutOfDomain. - A
riskySensitiveActionBlockedentry for the account in the same period shows that Google stopped one such attempt. Treat it as the same pattern.
Eventsgoogle.login.LoginService.suspiciousLogingoogle.login.LoginService.loginSuccessgoogle.login.LoginService.2svDisablegoogle.login.LoginService.passwordEditgoogle.login.LoginService.recoveryEmailEditgoogle.login.LoginService.recoveryPhoneEditgoogle.login.LoginService.emailForwardingOutOfDomaingoogle.login.LoginService.riskySensitiveActionBlocked
LimitsA user who is warned about a sign-in and then changes the password on their own produces the same sequence, so confirm with the user who made the change. The 24 hours are a starting point. Google's page shows how string parameters appear in protoPayload.metadata but not boolean ones, so check in your own logs how is_suspicious is written before filtering on it. All of these events are Data Access entries of login.googleapis.com.
Triage
- Read the affected account in the event parameters (affected_email_address in protoPayload.metadata). The suspicious login events and the suspension event also carry login_timestamp, the time of the sign-in that Google flagged. For a government-backed attack warning the user is the actor (principalEmail).
- Check what the account can reach, starting with admin roles in Workspace and IAM roles in Google Cloud.
- Review the account's other login entries around that time, such as successful sign-ins from new addresses, and ask the user through a second channel whether the attempt was theirs. For a programmatic sign-in, Google recommends finding out which application the user was using.
- Look for changes that follow a takeover, such as 2-Step Verification disabled, a changed password, changed recovery email or phone and mail forwarding out of the domain. The Login audit log has an event for each.
- If the user does not recognise the sign-in, treat the password as stolen even when the attempt was blocked.
Response
Once the attack is confirmed:
- Suspend the user until the account is secured, which is what Google recommends for a suspicious sign-in. Leave an account that Google suspended in that state until the steps below are done.
- Reset the password and reset the user's sign-in cookies, so that sessions opened with the old password end.
- Remove what a takeover leaves behind. Check mail forwarding and filters, recovery email and phone, 2-Step Verification methods, application passwords and third-party applications with access to the account.
- Review what the account did since the flagged sign-in, in Workspace and in Google Cloud, and undo changes nobody can explain. An admin account calls for a review of role assignments, domain-wide delegation and security settings.
- Restore the account only with 2-Step Verification enrolled, and have the user change the same password wherever else it was used.
Mitigation
- Enforce 2-Step Verification for every organizational unit and enforce security keys, as Google recommends in its guidance for these alerts.
- Move users and devices to applications that sign in with OAuth. Google Workspace accounts have not supported less secure apps since 1 May 2025, and the setting is no longer in the Admin console.
- Share Workspace data with Google Cloud and route the organization-level audit logs to the SIEM, including the Data Access log that carries the login events.
- Send the Alert Center notifications for these warnings to the security team, as a second path that does not depend on log routing.
Frequently asked questions
If Google blocked the suspicious sign-in, why does it still need attention?
Because the block does not change what the attacker knows. Google describes the alert as a sign-in attempt that did not match the user's normal behavior, raised in most cases after the user failed or abandoned a login challenge. Whoever made the attempt very likely has the password, and can try again from another place or use it on other services. Confirm with the user, and reset the password if the attempt was not theirs.
Why do these rules need the Data Access audit log?
Google Workspace Login Audit writes Data Access audit logs only. The entries appear in Cloud Audit Logs at organization level under the service name `login.googleapis.com`, and only when the organization shares Workspace data with Google Cloud. A SIEM that receives just the Admin Activity logs of the organization, or only the logs of projects, never sees these events.
What is the difference between the leaked password event and the suspension event?
`account_disabled_password_leak` means Google has become aware that someone else knows the password, and it requires a password reset before the user can sign in again. `account_disabled_hijacked` means Google detected activity suggesting that the account was already compromised and suspended it. The first is an exposed credential, the second a likely takeover, which is why its rule has the higher level.
Does a government-backed attack warning mean the account was compromised?
No. Google describes it as a warning about a potential government-backed attack, for example an attempt to steal a user's password, and the audit message says the user might have been targeted. It tells you who is being targeted, not that the attempt worked. Google's advice is to reset the passwords of affected users and to enforce 2-Step Verification and security keys.
References
- MITRE ATT&CK T1078.004: Valid Accounts: Cloud Accounts
- Audit logs for Google Workspace (Cloud Audit Logs)
- Configure audit logs for Google Workspace
- Login audit events (account warning and attack warning events)
- Protect Google Workspace accounts with security challenges
- View alert details (leaked password, suspicious login, user suspended, government-backed attacks)
- Control access to less secure apps