Detection rules › Panther

Databricks Account Admin Privileged Role Assignment

Status
Experimental
Severity
medium
Log types
Databricks.Audit
Tags
Databricks, Privilege Escalation, Persistence
Reference
https://github.com/databricks-solutions/cybersec-workspace-detection-app/blob/main/base/detections/event-based/account_admin_privileged_role_assignment.py
Source
github.com/panther-labs/panther-analysis

Detects when account-level admin privileges are granted in Databricks through direct role assignments or administrative group membership. Account admins have extensive control across all workspaces and should be carefully monitored. Successful grants are elevated to HIGH severity.

MITRE ATT&CK coverage

TacticTechniques
PersistenceT1098 Account Manipulation, T1136 Create Account
Privilege EscalationT1098 Account Manipulation

Rule body yaml

AnalysisType: rule
Filename: databricks_account_admin_privileged_role_assignment.py
RuleID: "Databricks.Audit.AccountAdminPrivilegedRoleAssignment"
DisplayName: "Databricks Account Admin Privileged Role Assignment"
Enabled: true
Status: Experimental
LogTypes:
  - Databricks.Audit
Tags:
  - Databricks
  - Privilege Escalation
  - Persistence
Reports:
  MITRE ATT&CK:
    - TA0004:T1098 # Account Manipulation
    - TA0003:T1136 # Create Account
Severity: Medium
Description: >
  Detects when account-level admin privileges are granted in Databricks through direct role
  assignments or administrative group membership. Account admins have extensive control across
  all workspaces and should be carefully monitored. Successful grants are elevated to HIGH severity.
Runbook: |
  1. Query audit logs for all account-level administrative actions by the target principal in the 24 hours after this privilege grant
  2. Check if the target principal accessed multiple workspaces or performed bulk operations in the 6 hours after receiving admin rights
  3. Find all account admin grants in the past 90 days to identify unusual patterns or privilege escalation chains
Reference: https://github.com/databricks-solutions/cybersec-workspace-detection-app/blob/main/base/detections/event-based/account_admin_privileged_role_assignment.py
SummaryAttributes:
  - actor
  - target_principal
  - principal_type
Tests:
  - Name: Set Account Admin Success
    ExpectedResult: true
    Log:
      timestamp: 1234567890000
      auditLevel: "ACCOUNT_LEVEL"
      serviceName: "accounts"
      actionName: "setAccountAdmin"
      accountId: "12345678-1234-1234-1234-123456789012"
      userIdentity:
        email: "superadmin@example.com"
      sourceIPAddress: "198.51.100.1"
      requestParams:
        targetUserName: "newadmin@example.com"
      response:
        statusCode: 200
  - Name: Change Account Owner
    ExpectedResult: true
    Log:
      timestamp: 1234567890000
      auditLevel: "ACCOUNT_LEVEL"
      serviceName: "accounts"
      actionName: "changeAccountOwner"
      userIdentity:
        email: "owner@example.com"
      requestParams:
        targetUserName: "newowner@example.com"
      response:
        statusCode: 200
  - Name: Add to Admin Group
    ExpectedResult: true
    Log:
      timestamp: 1234567890000
      auditLevel: "ACCOUNT_LEVEL"
      serviceName: "accounts"
      actionName: "addPrincipalToGroup"
      userIdentity:
        email: "admin@example.com"
      requestParams:
        principal: "user@example.com"
        targetGroupName: "account-admins"
      response:
        statusCode: 200
  - Name: Failed Admin Grant
    ExpectedResult: true
    Log:
      timestamp: 1234567890000
      auditLevel: "ACCOUNT_LEVEL"
      serviceName: "accounts"
      actionName: "setAccountAdmin"
      userIdentity:
        email: "attacker@example.com"
      requestParams:
        targetUserName: "attacker@example.com"
      response:
        statusCode: 403
  - Name: Workspace-Level Admin Grant
    ExpectedResult: false
    Log:
      timestamp: 1234567890000
      auditLevel: "WORKSPACE_LEVEL"
      serviceName: "accounts"
      actionName: "setAdmin"
      userIdentity:
        email: "admin@example.com"
      requestParams:
        targetUserName: "user@example.com"
  - Name: Remove Admin Should Not Alert
    ExpectedResult: false
    Log:
      timestamp: 1234567890000
      auditLevel: "ACCOUNT_LEVEL"
      serviceName: "accounts"
      actionName: "removeAdmin"
      userIdentity:
        email: "admin@example.com"
      requestParams:
        targetUserName: "user@example.com"
      response:
        statusCode: 200
  - Name: Remove From Admin Group Should Not Alert
    ExpectedResult: false
    Log:
      timestamp: 1234567890000
      auditLevel: "ACCOUNT_LEVEL"
      serviceName: "accounts"
      actionName: "removePrincipalFromGroup"
      userIdentity:
        email: "admin@example.com"
      requestParams:
        principal: "user@example.com"
        targetGroupName: "account-admins"
      response:
        statusCode: 200
  - Name: Non-Admin Group
    ExpectedResult: false
    Log:
      timestamp: 1234567890000
      auditLevel: "ACCOUNT_LEVEL"
      serviceName: "accounts"
      actionName: "addPrincipalToGroup"
      userIdentity:
        email: "admin@example.com"
      requestParams:
        principal: "user@example.com"
        targetGroupName: "developers"

Detection logic

Condition

auditLevel eq "ACCOUNT_LEVEL"
actionName not in ["removeAdmin", "removePrincipalFromGroup"]

This rule also runs imperative logic the parser cannot express as a filter; the conditions above are the structured part it could extract.

Exclusions

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

FieldKindExcluded values
actionNameinremoveAdmin, removePrincipalFromGroup

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
auditLeveleq
  • ACCOUNT_LEVEL

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
emailuserIdentity.email