How the attack works
Recovery in Azure rests on a few resource types. Azure Backup keeps recovery points of protected items in Recovery Services vaults and Backup vaults. Managed disk snapshots and VM restore points are point-in-time copies of virtual machine disks. Immutability policies on blob containers protect blobs against deletes and overwrites. Each of them can be removed with a control-plane operation: Microsoft.RecoveryServices/Vaults/backupFabrics/protectionContainers/protectedItems/delete and .../protectionContainers/delete for backup items and their containers, Microsoft.Compute/snapshots/delete and Microsoft.Compute/restorePointCollections/delete for disk copies, and Microsoft.Storage/storageAccounts/blobServices/containers/immutabilityPolicies/delete for an immutability policy.
Microsoft's report on Storm-0501 describes all of these in one incident. After the actor had made itself Owner of the subscriptions, it deleted snapshots, restore point collections, storage accounts and Recovery Services protection containers in a mass deletion that kept the victim from restoring data. Where resource locks and immutability policies made deletions fail, the actor deleted the locks and the policies and continued. Every one of these requests is recorded in the Azure Activity log with its caller. Single deletions of snapshots and restore points are routine, so the rule for them counts deletions per caller, and a second threshold rule counts how many different kinds of protection one caller removed. A deleted backup item, vault or immutability policy is a single event worth a review.
What an attacker needs
- A role with the delete actions on the recovery resources: Owner or Contributor on the scope, Backup Contributor for backup items, or a custom role with the operations named above.
- For an immutability policy: a policy that is still unlocked. Microsoft documents that a locked time-based retention policy cannot be deleted.
- For a vault that is protected by multi-user authorization: permissions on its Resource Guard as well. Microsoft defines disabling soft delete and removing the multi-user authorization as critical operations that the Resource Guard protects.
Impact
The organization loses the copies it would restore from: backup recovery points, point-in-time disk copies and the guarantee that blobs cannot be deleted or overwritten. On its own this destroys no production data, but it turns a later deletion or encryption into a permanent loss, which is the leverage for extortion. Backup data that is covered by soft delete stays recoverable for the retention period, 14 days by default, so the window to react is limited.
Detection rules
Azure Backup Item, Container, Vault or Resource Guard Link Deleted
Detects a successful deletion in Azure Backup that takes protection away: a protected item or a registered container of a Recovery Services vault, a backup instance of a Backup vault, a vault itself, or the Resource Guard proxy that links a vault to multi-user authorization. The report by Microsoft on Storm-0501 lists the deletion of Recovery Services protection containers among the operations of the actor's mass deletion. With soft delete the backup data stays recoverable for the retention period, which is 14 days by default, so the event leaves a limited time to react. Deleting the Resource Guard proxy removes the second authorization that protects critical vault operations.
title: Azure Backup Item, Container, Vault or Resource Guard Link Deleted
id: da67de73-c4b7-4b60-8408-e8f69a4d8f7e
status: experimental
description: >-
Detects a successful deletion in Azure Backup that takes protection away: a
protected item or a registered container of a Recovery Services vault, a
backup instance of a Backup vault, a vault itself, or the Resource Guard
proxy that links a vault to multi-user authorization. The report by
Microsoft on Storm-0501 lists the deletion of Recovery Services protection
containers among the operations of the actor's mass deletion. With
soft delete the backup data stays recoverable for the retention period,
which is 14 days by default, so the event leaves a limited time to react.
Deleting the Resource Guard proxy removes the second authorization that
protects critical vault operations.
references:
- https://attack.mitre.org/techniques/T1490/
- https://learn.microsoft.com/en-us/azure/backup/backup-azure-security-feature-cloud
- https://learn.microsoft.com/en-us/azure/backup/multi-user-authorization-concept
- https://learn.microsoft.com/en-us/azure/role-based-access-control/permissions/management-and-governance
- https://learn.microsoft.com/en-us/azure/role-based-access-control/permissions/security
- https://www.microsoft.com/en-us/security/blog/2025/08/27/storm-0501s-evolving-techniques-lead-to-cloud-based-ransomware/
author: Forestall
date: 2026-11-18
tags:
- attack.impact
- attack.t1490
logsource:
product: azure
service: activitylogs
detection:
selection:
OperationNameValue:
- MICROSOFT.RECOVERYSERVICES/VAULTS/BACKUPFABRICS/PROTECTIONCONTAINERS/PROTECTEDITEMS/DELETE
- MICROSOFT.RECOVERYSERVICES/VAULTS/BACKUPFABRICS/PROTECTIONCONTAINERS/DELETE
- MICROSOFT.RECOVERYSERVICES/VAULTS/BACKUPRESOURCEGUARDPROXIES/DELETE
- MICROSOFT.RECOVERYSERVICES/VAULTS/DELETE
- MICROSOFT.DATAPROTECTION/BACKUPVAULTS/BACKUPINSTANCES/DELETE
- MICROSOFT.DATAPROTECTION/BACKUPVAULTS/BACKUPRESOURCEGUARDPROXIES/DELETE
- MICROSOFT.DATAPROTECTION/BACKUPVAULTS/DELETE
ActivityStatusValue:
- Success
- Succeeded
condition: selection
falsepositives:
- Decommissioning of a workload, where its backup is stopped and deleted on purpose, and the clean-up of test vaults. Correlate with change records.
- Infrastructure-as-code pipelines that tear down non-production environments. Exclude their identities in a filter_ selection by Caller and keep alerting on production subscriptions.
level: medium
False positives
- Decommissioning of a workload, where its backup is stopped and deleted on purpose, and the clean-up of test vaults. Correlate with change records.
- Infrastructure-as-code pipelines that tear down non-production environments. Exclude their identities in a filter_ selection by Caller and keep alerting on production subscriptions.
Azure Blob Container Immutability Policy Deleted
Detects a successful deletion of the immutability policy of a blob container (Microsoft.Storage/storageAccounts/blobServices/containers/immutabilityPolicies/delete). An immutability policy protects the blobs of the container against deletes and overwrites. Microsoft documents that an unlocked time-based retention policy can be deleted and a locked one cannot, so the event means that an unlocked policy was removed and the data it covered can be deleted or overwritten again. The report by Microsoft on Storm-0501 describes the actor removing immutability policies with this operation in order to delete storage data.
title: Azure Blob Container Immutability Policy Deleted
id: b10c07e9-2d1a-44a2-b238-e9bb7f2ce679
status: experimental
description: >-
Detects a successful deletion of the immutability policy of a blob container
(Microsoft.Storage/storageAccounts/blobServices/containers/immutabilityPolicies/delete).
An immutability policy protects the blobs of the container against deletes
and overwrites. Microsoft documents that an unlocked time-based retention
policy can be deleted and a locked one cannot, so the event means that an
unlocked policy was removed and the data it covered can be deleted or
overwritten again. The report by Microsoft on Storm-0501 describes the actor
removing immutability policies with this operation in order to delete
storage data.
references:
- https://attack.mitre.org/techniques/T1490/
- https://learn.microsoft.com/en-us/azure/storage/blobs/immutable-time-based-retention-policy-overview
- https://learn.microsoft.com/en-us/azure/storage/blobs/immutable-policy-configure-container-scope
- https://learn.microsoft.com/en-us/azure/role-based-access-control/permissions/storage
- https://www.microsoft.com/en-us/security/blog/2025/08/27/storm-0501s-evolving-techniques-lead-to-cloud-based-ransomware/
author: Forestall
date: 2026-11-18
tags:
- attack.impact
- attack.t1490
logsource:
product: azure
service: activitylogs
detection:
selection:
OperationNameValue: MICROSOFT.STORAGE/STORAGEACCOUNTS/BLOBSERVICES/CONTAINERS/IMMUTABILITYPOLICIES/DELETE
ActivityStatusValue:
- Success
- Succeeded
condition: selection
falsepositives:
- Administrators who remove an unlocked policy while they test retention settings, before the policy is locked. Correlate with change records.
level: medium
False positives
- Administrators who remove an unlocked policy while they test retention settings, before the policy is locked. Correlate with change records.
Many Azure Snapshots or VM Restore Points Deleted by One Caller
Counts, per Caller, the successful deletions of managed disk snapshots, VM restore points and restore point collections. Ten or more within 30 minutes is the pattern of somebody clearing the point-in-time copies of virtual machine disks, which the report by Microsoft on Storm-0501 describes as part of the mass deletion that kept the victim from restoring data. A single deletion is routine, which is why the rule counts. The count and the window are starting points to tune.
Fires when one Caller has at least 10 matching events within 30 minutes.
title: Azure Snapshot or VM Restore Point Deleted (Threshold Base Event)
id: a289777b-d7bc-47de-97ea-8492e4ef55dc
name: azure_snapshot_or_restore_point_deleted
status: experimental
description: >-
Base event of the threshold rule in this file. A successful deletion of a
managed disk snapshot, a VM restore point or a restore point collection in
the Azure Activity log.
logsource:
product: azure
service: activitylogs
detection:
selection:
OperationNameValue:
- MICROSOFT.COMPUTE/SNAPSHOTS/DELETE
- MICROSOFT.COMPUTE/RESTOREPOINTCOLLECTIONS/DELETE
- MICROSOFT.COMPUTE/RESTOREPOINTCOLLECTIONS/RESTOREPOINTS/DELETE
ActivityStatusValue:
- Success
- Succeeded
condition: selection
level: low
---
title: Many Azure Snapshots or VM Restore Points Deleted by One Caller
id: 0d284a2b-0767-48f2-b403-210978bc2d3e
status: experimental
description: >-
Counts, per Caller, the successful deletions of managed disk snapshots, VM
restore points and restore point collections. Ten or more within 30 minutes
is the pattern of somebody clearing the point-in-time copies of virtual
machine disks, which the report by Microsoft on Storm-0501 describes as part
of the mass deletion that kept the victim from restoring data. A single
deletion is routine, which is why the rule counts. The count and the window
are starting points to tune.
references:
- https://attack.mitre.org/techniques/T1490/
- https://learn.microsoft.com/en-us/azure/role-based-access-control/permissions/compute
- https://learn.microsoft.com/en-us/azure/virtual-machines/virtual-machines-create-restore-points
- https://www.microsoft.com/en-us/security/blog/2025/08/27/storm-0501s-evolving-techniques-lead-to-cloud-based-ransomware/
author: Forestall
date: 2026-11-18
tags:
- attack.impact
- attack.t1490
correlation:
type: event_count
rules:
- azure_snapshot_or_restore_point_deleted
group-by:
- Caller
timespan: 30m
condition:
gte: 10
falsepositives:
- Backup and disaster recovery products that prune expired snapshots and restore points in batches. Exclude their identities by Caller, or raise the count for them.
- Clean-up after a migration or an image-building run, and the teardown of test environments. Correlate with change records.
- The values of 10 deletions and 30 minutes are starting points, not tested limits. Measure what your own backup jobs produce and adjust both.
level: medium
False positives
- Backup and disaster recovery products that prune expired snapshots and restore points in batches. Exclude their identities by Caller, or raise the count for them.
- Clean-up after a migration or an image-building run, and the teardown of test environments. Correlate with change records.
- The values of 10 deletions and 30 minutes are starting points, not tested limits. Measure what your own backup jobs produce and adjust both.
Azure Recovery Services Vault Backup Configuration Changed
Detects a successful update of the backup configuration of a Recovery Services vault (Microsoft.RecoveryServices/Vaults/backupconfig/write). This is the resource that holds the soft delete state and the enhanced security state of the vault, so the operation is how those protections are changed. The Activity log event does not say which property was changed or to which value, and the rule matches every update of the configuration, so read the current security settings of the vault and use the rule for review and correlation with the deletion rules on this page.
title: Azure Recovery Services Vault Backup Configuration Changed
id: 84beb3d7-722e-4e77-baf3-98ce23c3877e
status: experimental
description: >-
Detects a successful update of the backup configuration of a Recovery
Services vault (Microsoft.RecoveryServices/Vaults/backupconfig/write). This
is the resource that holds the soft delete state and the enhanced security
state of the vault, so the operation is how those protections are changed.
The Activity log event does not say which property was changed or to which
value, and the rule matches every update of the configuration, so read the
current security settings of the vault and use the rule for review and
correlation with the deletion rules on this page.
references:
- https://attack.mitre.org/techniques/T1490/
- https://learn.microsoft.com/en-us/rest/api/backup/backup-resource-vault-configs/update
- https://learn.microsoft.com/en-us/azure/backup/backup-azure-security-feature-cloud
- https://learn.microsoft.com/en-us/azure/role-based-access-control/permissions/management-and-governance
author: Forestall
date: 2026-11-18
tags:
- attack.impact
- attack.t1490
logsource:
product: azure
service: activitylogs
detection:
selection:
OperationNameValue: MICROSOFT.RECOVERYSERVICES/VAULTS/BACKUPCONFIG/WRITE
ActivityStatusValue:
- Success
- Succeeded
condition: selection
falsepositives:
- Backup administrators who change vault settings, and templates that set the configuration on every deployment. Exclude the deployment identities in a filter_ selection by Caller once confirmed.
level: low
False positives
- Backup administrators who change vault settings, and templates that set the configuration on every deployment. Exclude the deployment identities in a filter_ selection by Caller once confirmed.
Several Kinds of Azure Recovery Protection Removed by One Caller
Counts, per Caller, how many different removal operations it completed among backup items, backup containers and vaults, Resource Guard proxies, disk snapshots, VM restore points, blob immutability policies and management locks. Three or more different operations within 4 hours is the pattern of somebody removing whatever stands in the way of destruction, as the report by Microsoft on Storm-0501 describes it, and not of routine work, where one kind of object is deleted for a reason that is on record. The count and the window are starting points to tune.
Fires when one Caller has at least 3 different OperationNameValue values within 4 hours.
title: Azure Recovery Protection Removed (Threshold Base Event)
id: e959349b-0bd6-4c11-bd61-257252f61772
name: azure_recovery_protection_removed
status: experimental
description: >-
Base event of the threshold rule in this file. A successful removal of a
recovery protection in the Azure Activity log: a backup item, container or
vault, a Resource Guard proxy, a managed disk snapshot, a VM restore point
or restore point collection, a blob container immutability policy or a
management lock.
logsource:
product: azure
service: activitylogs
detection:
selection:
OperationNameValue:
- MICROSOFT.RECOVERYSERVICES/VAULTS/BACKUPFABRICS/PROTECTIONCONTAINERS/PROTECTEDITEMS/DELETE
- MICROSOFT.RECOVERYSERVICES/VAULTS/BACKUPFABRICS/PROTECTIONCONTAINERS/DELETE
- MICROSOFT.RECOVERYSERVICES/VAULTS/BACKUPRESOURCEGUARDPROXIES/DELETE
- MICROSOFT.RECOVERYSERVICES/VAULTS/DELETE
- MICROSOFT.DATAPROTECTION/BACKUPVAULTS/BACKUPINSTANCES/DELETE
- MICROSOFT.DATAPROTECTION/BACKUPVAULTS/BACKUPRESOURCEGUARDPROXIES/DELETE
- MICROSOFT.DATAPROTECTION/BACKUPVAULTS/DELETE
- MICROSOFT.COMPUTE/SNAPSHOTS/DELETE
- MICROSOFT.COMPUTE/RESTOREPOINTCOLLECTIONS/DELETE
- MICROSOFT.COMPUTE/RESTOREPOINTCOLLECTIONS/RESTOREPOINTS/DELETE
- MICROSOFT.STORAGE/STORAGEACCOUNTS/BLOBSERVICES/CONTAINERS/IMMUTABILITYPOLICIES/DELETE
- MICROSOFT.AUTHORIZATION/LOCKS/DELETE
ActivityStatusValue:
- Success
- Succeeded
condition: selection
level: low
---
title: Several Kinds of Azure Recovery Protection Removed by One Caller
id: fb480eee-3cb9-4482-aefc-eed351ea81a9
status: experimental
description: >-
Counts, per Caller, how many different removal operations it completed among
backup items, backup containers and vaults, Resource Guard proxies, disk
snapshots, VM restore points, blob immutability policies and management
locks. Three or more different operations within 4 hours is the pattern of
somebody removing whatever stands in the way of destruction, as the report
by Microsoft on Storm-0501 describes it, and not of routine work, where one
kind of object is deleted for a reason that is on record. The count and the
window are starting points to tune.
references:
- https://attack.mitre.org/techniques/T1490/
- https://learn.microsoft.com/en-us/azure/role-based-access-control/permissions/management-and-governance
- https://learn.microsoft.com/en-us/azure/role-based-access-control/permissions/compute
- https://learn.microsoft.com/en-us/azure/role-based-access-control/permissions/storage
- https://learn.microsoft.com/en-us/azure/role-based-access-control/permissions/security
- https://www.microsoft.com/en-us/security/blog/2025/08/27/storm-0501s-evolving-techniques-lead-to-cloud-based-ransomware/
author: Forestall
date: 2026-11-18
tags:
- attack.impact
- attack.t1490
correlation:
type: value_count
rules:
- azure_recovery_protection_removed
group-by:
- Caller
timespan: 4h
condition:
field: OperationNameValue
gte: 3
falsepositives:
- The teardown of an environment, which removes its backups, snapshots and locks together. Correlate with change records and exclude the pipeline identities by Caller.
- Backup administrators during a migration between vaults. The values of three operations and 4 hours are starting points, not tested limits.
level: high
False positives
- The teardown of an environment, which removes its backups, snapshots and locks together. Correlate with change records and exclude the pipeline identities by Caller.
- Backup administrators during a migration between vaults. The values of three operations and 4 hours are starting points, not tested limits.
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.
Protection removed, then the protected data deleted by the same caller
The removal of a backup or an immutability policy becomes an incident when the data behind it is deleted next. Microsoft describes this order for Storm-0501, where storage accounts were deleted after their protections. Joining the two by caller and resource shows which data is now gone without a copy.
- In the Azure Activity log, take a successful
MICROSOFT.STORAGE/STORAGEACCOUNTS/BLOBSERVICES/CONTAINERS/IMMUTABILITYPOLICIES/DELETEor one of the backup deletions of this page. Keep theCallerand the resource ID. - Within 24 hours, look for
MICROSOFT.STORAGE/STORAGEACCOUNTS/DELETEorMICROSOFT.COMPUTE/DISKS/DELETEwith status Success by the sameCaller, for the storage account of the policy or in the same resource group. - List the affected resources and check for each whether a backup, a snapshot or soft-deleted data still exists.
EventsMICROSOFT.STORAGE/STORAGEACCOUNTS/BLOBSERVICES/CONTAINERS/IMMUTABILITYPOLICIES/DELETEMICROSOFT.STORAGE/STORAGEACCOUNTS/DELETEMICROSOFT.COMPUTE/DISKS/DELETE
LimitsDecommissioning a workload deletes its backup and then its data in the same order. Blobs that are deleted through the storage service itself are data-plane requests and are not in the Activity log. The 24 hours are a starting point.
Triage
- Identify the Caller, the caller's IP address and the resources from the events, and check for a decommissioning or clean-up that explains them.
- Decide whether the caller is a backup or disaster recovery service that prunes its own snapshots and restore points, or a person or an identity that never did this before.
- List what the same caller removed in the same period across types, such as backup items, vaults, snapshots, restore points, immutability policies, resource locks and storage accounts. Several types together are the pattern of preparation for destruction.
- For a backup deletion, check whether soft delete was enabled on the vault and how long the deleted data stays recoverable.
- Review how the caller holds its role and whether the role was assigned or activated shortly before.
Response
Once the attack is confirmed:
- Contain the caller first. Revoke its sessions, rotate its credentials and remove its Owner, Contributor and backup roles, so that the deletion stops.
- Recover what is still recoverable before the retention ends. Undelete soft-deleted backup items and vaults, and recover soft-deleted storage data where soft delete was enabled.
- Put back the protections that were removed, such as immutability policies, resource locks and the Resource Guard association of the vaults, and lock the immutability policies this time.
- Take new backups or snapshots of the workloads that lost their copies, into a vault or scope the caller could not reach.
- Review the Activity log for deletions of production data by the same caller, and for the role assignments that gave it access, and remove those.
- Check the vault security settings, such as soft delete and immutability, against their expected state.
Mitigation
- Keep soft delete enabled on Recovery Services vaults and Backup vaults, and protect critical vault operations with multi-user authorization through a Resource Guard that the vault administrators do not control.
- Enable the immutable vault setting on backup vaults and lock it once you are satisfied with its effect, so that it cannot be disabled.
- Lock time-based retention policies on blob containers that hold data which must survive. An unlocked policy can be deleted by anyone with the storage account rights.
- Limit Owner, Contributor and Backup Contributor on the subscriptions and resource groups that hold vaults, snapshots and protected storage accounts, and assign them through Privileged Identity Management.
- Keep a copy of critical backups in a separate subscription or tenant that the administrators of the workload cannot reach.
- Send the Azure Activity log to your SIEM and alert on the operations of this page.
Frequently asked questions
Can deleted backup data be recovered?
Often, for a limited time. Microsoft documents that soft delete in Azure Backup keeps deleted backup data for a retention period of 14 days by default, which can be extended up to 180 days, and that the data can be restored while it is in the soft-deleted state. Deleted managed disk snapshots and VM restore points have no such safety net in the documentation these rules are based on, so treat them as gone.
Why are two rules thresholds and the others single events?
Snapshots and restore points are created and deleted all the time by backup products, image pipelines and administrators, so one deletion says nothing. Ten or more by one caller within 30 minutes is a starting point for the pattern of clearing copies. The second threshold counts different kinds of removal by one caller, because three kinds within 4 hours are unusual even when each single event has an explanation. A deleted backup item, vault or immutability policy is also reported one by one.
Does an immutability policy stop an attacker with the Owner role?
Only when it is locked. Microsoft documents that both locked and unlocked policies protect against deletes and overwrites, but that an unlocked policy can be shortened or deleted, while a locked time-based retention policy cannot be deleted and its retention cannot be decreased. The rule on this page therefore sees the removal of unlocked policies.
Do these rules cover the deletion of the data itself?
No. They cover the removal of backups, copies and protections, which is the step before. Deleting a key vault is a separate attack in this library. Deleting storage accounts, deleting blobs through the storage service and encrypting data with a new key are not matched by the rules on this page.
References
- MITRE ATT&CK T1490: Inhibit System Recovery
- Secure by default with soft delete for Azure Backup
- Multiuser authorization using Resource Guard
- Immutable vault for Azure Backup
- Time-based retention policies for immutable blob data
- Overview of VM restore points
- Azure permissions for Management and governance
- Azure permissions for Compute
- Azure permissions for Storage
- Azure permissions for Security
- Microsoft Security Blog: Storm-0501's evolving techniques lead to cloud-based ransomware