Detect Password Spray Attack Behavior From Source

name: Detect Password Spray Attack Behavior From Source
id: b6391b15-e913-4c2c-8949-9eecc06efacc
version: 12
creation_date: '2023-11-11'
modification_date: '2026-05-13'
author: Steven Dick
status: production
type: TTP
description: The following analytic identifies one source failing to authenticate with 10 or more unique users. This behavior could represent an adversary performing a Password Spraying attack to obtain initial access or elevate privileges. This logic can be used for real time security monitoring as well as threat hunting exercises and works well against any number of data sources ingested into the CIM datamodel. Environments can be very different depending on the organization. Test and customize this detections thresholds if needed.
data_source:
    - Windows Event Log Security 4624
    - Windows Event Log Security 4625
search: |-
    | tstats `security_content_summariesonly` max(_time) as lastTime, min(_time) as firstTime, values(Authentication.user_category) as user_category values(Authentication.src_category) as src_category values(Authentication.app) as app count FROM datamodel=Authentication.Authentication
      BY Authentication.action Authentication.app Authentication.authentication_method
         Authentication.dest Authentication.signature Authentication.signature_id
         Authentication.src Authentication.user
    | `drop_dm_object_name("Authentication")`
    | eval user=case((match(upper(user),"[a-zA-Z0-9]{3}")),upper(user),true(),null), src=upper(src), success=if(action="success",count,0),success_user=if(action="success",user,null),failure=if(action="failure",count,0), failed_user=if(action="failure",user,null)
    | stats count min(firstTime) as firstTime max(lastTime) as lastTime values(app) as app values(src_category) as src_category values(success_user) as user values(failed_user) as failed_user dc(success_user) as success_dc dc(failed_user) as failed_dc dc(user) as user_dc ,sum(failure) as failure,sum(success) as success
      BY src
    | fields - _time
    | where user_dc >= 10 AND .25 > (success/failure) AND failed_dc > success_dc
    | `security_content_ctime(firstTime)`
    | `security_content_ctime(lastTime)`
    | `detect_password_spray_attack_behavior_from_source_filter`
how_to_implement: This detection requires ingesting authentication data to the appropriate accelerated datamodel. Recommend adjusting the search time window for this correlation to match the number of unique users (user_dc) in hours. i.e. 10 users over 10hrs
known_false_positives: Domain controllers, authentication chokepoints, and vulnerability scanners.
references:
    - https://attack.mitre.org/techniques/T1110/003/
    - https://www.microsoft.com/en-us/security/blog/2020/04/23/protecting-organization-password-spray-attacks/
    - https://github.com/MarkoH17/Spray365
finding:
    title: The source [$src$] attempted to access $user_dc$ distinct users a total of $count$ times between [$firstTime$] and [$lastTime$]. $success$ successful logins detected.
    entity:
        field: user
        type: user
        score: 50
intermediate_findings:
    entities:
        - field: src
          type: system
          score: 50
          message: The source [$src$] attempted to access $user_dc$ distinct users a total of $count$ times between [$firstTime$] and [$lastTime$]. $success$ successful logins detected.
analytic_story:
    - Compromised User Account
asset_type: Account
mitre_attack_id:
    - T1110.003
product:
    - Splunk Enterprise
    - Splunk Enterprise Security
    - Splunk Cloud
category: endpoint
security_domain: access