Detection rules › Kusto

Suspicious VM Instance Creation Activity Detected

Severity
medium
Time window
1d
Group by
AccountObjectId, AlertName, AlertSeverity, AlertTime, EntityIp, GCPUserIp, ProductName, ProviderName, Tactics, Techniques
Source
github.com/Azure/Azure-Sentinel

'This detection identifies high-severity alerts across various Microsoft security products, including Microsoft Defender XDR and Microsoft Entra ID, and correlates them with instances of Google Cloud VM creation. It focuses on instances where VMs were created within a short timeframe of high-severity alerts, potentially indicating suspicious activity.'

MITRE ATT&CK coverage

TacticTechniques
Initial AccessT1078 Valid Accounts
ExecutionT1106 Native API
DiscoveryT1526 Cloud Service Discovery

Rules detecting the same action

Other rules on this platform that filter on the same API call or operation.

Rule body kusto

id: 1cc0ba27-c5ca-411a-a779-fbc89e26be83
name: Suspicious VM Instance Creation Activity Detected
description: |
  'This detection identifies high-severity alerts across various Microsoft security products, including Microsoft Defender XDR and Microsoft Entra ID, and correlates them with instances of Google Cloud VM creation. It focuses on instances where VMs were created within a short timeframe of high-severity alerts, potentially indicating suspicious activity.'
severity: Medium
requiredDataConnectors:
  - connectorId: GCPAuditLogsDefinition
    dataTypes:
      - GCPAuditLogs
  - connectorId: AzureActiveDirectoryIdentityProtection
    dataTypes:
      - SecurityAlert (IPC)
  - connectorId: MicrosoftThreatProtection
    dataTypes:
      - SecurityAlert
  - connectorId: MicrosoftDefenderAdvancedThreatProtection
    dataTypes:
      - SecurityAlert (MDATP)
  - connectorId: MicrosoftCloudAppSecurity
    dataTypes:
      - SecurityAlert
  - connectorId: BehaviorAnalytics
    dataTypes:
      - IdentityInfo
queryFrequency: 1d
queryPeriod: 1d
triggerOperator: gt
triggerThreshold: 0
tactics:
  - InitialAccess
  - Execution
  - Discovery
relevantTechniques:
  - T1078
  - T1106
  - T1526
query: |
    // Filter alerts from specific Microsoft security products with medium and high severity
    SecurityAlert 
    | where ProductName in ("Microsoft 365 Defender", "Azure Active Directory", "Microsoft Defender Advanced Threat Protection", "Microsoft Cloud App Security", "Azure Active Directory Identity Protection", "Microsoft Defender ATP")
    | where AlertSeverity has_any ("Medium", "High")
    // Parse JSON entities and extend AlertTimeGenerated
    | extend Entities = parse_json(Entities), AlertTimeGenerated=TimeGenerated
    // Extract and process IP entities
    | mv-apply Entity = Entities on 
        ( 
        where Entity.Type == 'ip' 
        | extend EntityIp = tostring(Entity.Address) 
        ) 
    // Extract and process account entities
    | mv-apply Entity = Entities on 
        ( 
        where Entity.Type == 'account' 
        | extend AccountObjectId = tostring(Entity.AadUserId)
        )
    // Filter out records with empty EntityIp
    | where isnotempty(EntityIp)
    // Summarize data and create sets of entities and system alert IDs
    | summarize Entitys=make_set(Entity), SystemAlertIds=make_set(SystemAlertId)
        by 
        AlertName,
        ProductName,
        AlertSeverity,
        EntityIp,
        Tactics,
        Techniques,
        ProviderName,
        AlertTime= bin(AlertTimeGenerated, 1d),
        AccountObjectId
    // Join with GCPAuditLogs for VM instance creation
    | join kind=inner (
        GCPAuditLogs
        | where ServiceName == "compute.googleapis.com" and MethodName endswith "instances.insert"
        | extend
            GCPUserUPN= tostring(parse_json(AuthenticationInfo).principalEmail),
            GCPUserIp = tostring(parse_json(RequestMetadata).callerIp),
            GCPUserUA= tostring(parse_json(RequestMetadata).callerSuppliedUserAgent),
            VMStatus =  tostring(parse_json(Response).status),
            VMOperation=tostring(parse_json(Response).operationType),
            VMName= tostring(parse_json(Request).name),
            VMType = tostring(split(parse_json(Request).machineType, "/")[-1])
        | where GCPUserUPN !has "gserviceaccount.com"
        | where VMOperation == "insert" and isnotempty(GCPUserIp) and GCPUserIp != "private"
        | project
            GCPOperationTime=TimeGenerated,
            VMName,
            VMStatus,
            MethodName,
            GCPUserUPN,
            ProjectId,
            GCPUserIp,
            GCPUserUA,
            VMOperation,
            VMType
        )
        on $left.EntityIp == $right.GCPUserIp 
    // Join with IdentityInfo to enrich user identity details
    | join kind=inner (IdentityInfo 
        | distinct AccountObjectId, AccountUPN, JobTitle
        )
        on AccountObjectId 
    // Calculate the time difference between the alert and VM creation for further analysis
    | extend TimeDiff= datetime_diff('day', AlertTime, GCPOperationTime),Name = split(GCPUserUPN, "@")[0], UPNSuffix = split(GCPUserUPN, "@")[1]
entityMappings:
  - entityType: IP
    fieldMappings:
      - identifier: Address
        columnName: GCPUserIp
  - entityType: Account
    fieldMappings:
      - identifier: FullName
        columnName: GCPUserUPN
      - identifier: Name
        columnName: Name
      - identifier: UPNSuffix
        columnName: UPNSuffix
customDetails:
  AlertName: AlertName
  AlertProDuctName: ProductName
  AlertUserName: AccountUPN
  AlertUserObjectId: AccountObjectId
  AlertIds: SystemAlertIds
  AlertIp: EntityIp
  GCPUserAgent: GCPUserUA
  GCPVMName: VMName
  GCPProjectId: ProjectId
  GCPVMType: VMType
  CorrelationWith: "GCPAuditLogs"
alertDetailsOverride:
  alertDisplayNameFormat: "IP address {{GCPUserIp}} Assocated with {{AlertName}} found in GCP VM creation event by {{GCPUserUPN}}"
  alertDescriptionFormat: "This detection correlates '{{ProductName}}' Alert IP addresse Entity found in VM instance creation in GCP {{ProjectId}}. It identifies successful compute instance creation, from suspicious IP addresse. By joining these datasets on network entities and IP addresses, it detects unauthorized Initial access attempts across GCP environments."
  alertSeverityColumnName: AlertSeverity
  alertDynamicProperties:
    - alertProperty: ProviderName
      value: "Microsoft Security"
    - alertProperty: ProductName
      value: "Microsoft Defender"
    - alertProperty: ProductComponentName
      value: "Microsoft Defender"
kind: Scheduled
version: 1.0.4

Stages and Predicates

Stage 1: source

SecurityAlert

Stage 2: where

| where ProductName in ("Microsoft 365 Defender", "Azure Active Directory", "Microsoft Defender Advanced Threat Protection", "Microsoft Cloud App Security", "Azure Active Directory Identity Protection", "Microsoft Defender ATP")

Stage 3: where

| where AlertSeverity has_any ("Medium", "High")

Stage 4: extend

| extend Entities = parse_json(Entities), AlertTimeGenerated=TimeGenerated

Stage 5: kusto:mv-apply

| mv-apply Entity = Entities on 
    ( 
    where Entity.Type == 'ip' 
    | extend EntityIp = tostring(Entity.Address) 
    )

Stage 6: kusto:mv-apply

| mv-apply Entity = Entities on 
    ( 
    where Entity.Type == 'account' 
    | extend AccountObjectId = tostring(Entity.AadUserId)
    )

Stage 7: where

| where isnotempty(EntityIp)

Stage 8: summarize

| summarize Entitys=make_set(Entity), SystemAlertIds=make_set(SystemAlertId)
    by 
    AlertName,
    ProductName,
    AlertSeverity,
    EntityIp,
    Tactics,
    Techniques,
    ProviderName,
    AlertTime= bin(AlertTimeGenerated, 1d),
    AccountObjectId

Stage 9: join

| join kind=inner (
    GCPAuditLogs
    | where ServiceName == "compute.googleapis.com" and MethodName endswith "instances.insert"
    | extend
        GCPUserUPN= tostring(parse_json(AuthenticationInfo).principalEmail),
        GCPUserIp = tostring(parse_json(RequestMetadata).callerIp),
        GCPUserUA= tostring(parse_json(RequestMetadata).callerSuppliedUserAgent),
        VMStatus =  tostring(parse_json(Response).status),
        VMOperation=tostring(parse_json(Response).operationType),
        VMName= tostring(parse_json(Request).name),
        VMType = tostring(split(parse_json(Request).machineType, "/")[-1])
    | where GCPUserUPN !has "gserviceaccount.com"
    | where VMOperation == "insert" and isnotempty(GCPUserIp) and GCPUserIp != "private"
    | project
        GCPOperationTime=TimeGenerated,
        VMName,
        VMStatus,
        MethodName,
        GCPUserUPN,
        ProjectId,
        GCPUserIp,
        GCPUserUA,
        VMOperation,
        VMType
    )
    on $left.EntityIp == $right.GCPUserIp

Stage 10: join

| join kind=inner (IdentityInfo 
    | distinct AccountObjectId, AccountUPN, JobTitle
    )
    on AccountObjectId

Stage 11: extend

| extend TimeDiff= datetime_diff('day', AlertTime, GCPOperationTime),Name = split(GCPUserUPN, "@")[0], UPNSuffix = split(GCPUserUPN, "@")[1]

Exclusions

Top-level NOT(...) conjuncts: predicates this rule actively suppresses.

FieldKindExcluded values
GCPUserUPNmatchgserviceaccount.com

Indicators

Each row is a field, operator, and value that the rule matches. The corpus column counts how many other rules in the catalog look for the same combination: high numbers point to widely-used, community-vetted indicators. Blank or 1 shows that the indicator is specific to this rule.

FieldKindValues
AlertSeveritymatch
  • High
  • Medium
EntityIpis_not_null
  • (no value, null check)
GCPUserIpis_not_null
  • (no value, null check)
GCPUserIpne
  • private transforms: cased
MethodNameends_with
  • instances.insert
ProductNamein
  • Azure Active Directory transforms: cased
  • Azure Active Directory Identity Protection transforms: cased
  • Microsoft 365 Defender transforms: cased
  • Microsoft Cloud App Security transforms: cased
  • Microsoft Defender ATP transforms: cased
  • Microsoft Defender Advanced Threat Protection transforms: cased
ServiceNameeq
  • compute.googleapis.com transforms: cased
Typeeq
  • account transforms: cased
  • ip transforms: cased
VMOperationeq
  • insert transforms: cased

Output fields

Fields the rule emits when it matches. Chronicle authors list these in the outcome block; they appear on the detection and $risk_score drives alerting. Sentinel / Defender XDR rules build them up through project / summarize / extend stages. Sentinel maps these into alert fields via entityMappings and customDetails; Defender XDR custom detections surface them as alert fields directly.

FieldSource
AccountObjectIdsummarize
AlertNamesummarize
AlertSeveritysummarize
AlertTimesummarize
EntityIpsummarize
Entityssummarize
ProductNamesummarize
ProviderNamesummarize
SystemAlertIdssummarize
Tacticssummarize
Techniquessummarize
Nameextend
TimeDiffextend
UPNSuffixextend