Detection rules › Splunk

Windows Unusual Count Of Invalid Users Failed To Auth Using NTLM

Status
production
Severity
low
Group by
_time, workstation_name
Author
Mauricio Velazco, Splunk
Source
github.com/splunk/security_content

The following analytic identifies a source endpoint failing to authenticate with multiple invalid users using the NTLM protocol. It leverages EventCode 4776 and calculates the standard deviation for each host, using the 3-sigma rule to detect anomalies. This behavior is significant as it may indicate a Password Spraying attack, where an adversary attempts to gain initial access or elevate privileges. If confirmed malicious, this activity could lead to unauthorized access or privilege escalation, posing a significant threat to the Active Directory environment. This detection is focused on domain controllers.

MITRE ATT&CK coverage

TacticTechniques
Credential AccessT1110.003 Brute Force: Password Spraying

Event coverage

Rule body splunk

name: Windows Unusual Count Of Invalid Users Failed To Auth Using NTLM
id: 15603165-147d-4a6e-9778-bd0ff39e668f
version: 12
creation_date: '2021-04-14'
modification_date: '2026-05-13'
author: Mauricio Velazco, Splunk
status: production
type: Anomaly
description: The following analytic identifies a source endpoint failing to authenticate with multiple invalid users using the NTLM protocol. It leverages EventCode 4776 and calculates the standard deviation for each host, using the 3-sigma rule to detect anomalies. This behavior is significant as it may indicate a Password Spraying attack, where an adversary attempts to gain initial access or elevate privileges. If confirmed malicious, this activity could lead to unauthorized access or privilege escalation, posing a significant threat to the Active Directory environment. This detection is focused on domain controllers.
data_source:
    - Windows Event Log Security 4776
search: |-
    `wineventlog_security` EventCode=4776 TargetUserName!=*$ Status=0xc0000064
      | bucket span=2m _time
      | stats dc(TargetUserName) AS unique_accounts values(TargetUserName) as user values(dest) as dest
        BY _time, Workstation
      | eventstats avg(unique_accounts) as comp_avg , stdev(unique_accounts) as comp_std
        BY Workstation
      | eval upperBound=(comp_avg+comp_std*3)
      | eval isOutlier=if(unique_accounts > 10 and unique_accounts >= upperBound, 1, 0)
      | search isOutlier=1
      | rename Workstation as src
      | `windows_unusual_count_of_invalid_users_failed_to_auth_using_ntlm_filter`
how_to_implement: To successfully implement this search, you need to be ingesting Domain Controller events. The Advanced Security Audit policy setting `Audit Credential Validation' within `Account Logon` needs to be enabled.
known_false_positives: A host failing to authenticate with multiple invalid domain users is not a common behavior for legitimate systems. Possible false positive scenarios include but are not limited to vulnerability scanners and missconfigured systems. If this detection triggers on a host other than a Domain Controller, the behavior could represent a password spraying attack against the host's local accounts.
references:
    - https://attack.mitre.org/techniques/T1110/003/
    - https://docs.microsoft.com/en-us/windows/security/threat-protection/auditing/audit-credential-validation
    - https://docs.microsoft.com/en-us/windows/security/threat-protection/auditing/event-4776
drilldown_searches:
    - name: View the detection results for - "$user$"
      search: '%original_detection_search% | search  user = "$user$"'
      earliest_offset: $info_min_time$
      latest_offset: $info_max_time$
    - name: View risk events for the last 7 days for - "$user$"
      search: '| from datamodel Risk.All_Risk | search normalized_risk_object IN ("$user$") | stats count min(_time) as firstTime max(_time) as lastTime values(search_name) as "Search Name" values(risk_message) as "Risk Message" values(analyticstories) as "Analytic Stories" values(annotations._all) as "Annotations" values(annotations.mitre_attack.mitre_tactic) as "ATT&CK Tactics" by normalized_risk_object | `security_content_ctime(firstTime)` | `security_content_ctime(lastTime)`'
      earliest_offset: 7d
      latest_offset: "0"
intermediate_findings:
    entities:
        - field: user
          type: user
          score: 20
          message: Potential NTLM based password spraying attack from $src$
        - field: src
          type: system
          score: 20
          message: Potential NTLM based password spraying attack from $src$
analytic_story:
    - Active Directory Password Spraying
    - Volt Typhoon
asset_type: Endpoint
mitre_attack_id:
    - T1110.003
product:
    - Splunk Enterprise
    - Splunk Enterprise Security
    - Splunk Cloud
category: endpoint
security_domain: endpoint
tests:
    - attack_data:
        - data: https://media.githubusercontent.com/media/splunk/attack_data/master/datasets/attack_techniques/T1110.003/purplesharp_invalid_users_ntlm_xml/windows-security.log
          source: XmlWinEventLog:Security
          sourcetype: XmlWinEventLog
      name: True Positive Test
      test_type: unit

Stages and Predicates

Stage 1: search

`wineventlog_security` EventCode=4776 TargetUserName!=*$ Status=0xc0000064

Stage 2: bucket

| bucket span=2m _time

Stage 3: stats

| stats dc(TargetUserName) AS unique_accounts values(TargetUserName) as user values(dest) as dest
    BY _time, Workstation

Stage 4: eventstats

| eventstats avg(unique_accounts) as comp_avg , stdev(unique_accounts) as comp_std
    BY Workstation

Stage 5: eval

| eval upperBound=(comp_avg+comp_std*3)

Stage 6: eval

| eval isOutlier=if(unique_accounts > 10 and unique_accounts >= upperBound, 1, 0)
isOutlier =
ifunique_accounts > 10 AND unique_accounts >= upperBound1
else0

Stage 7: search

| search isOutlier=1

Stage 8: rename

| rename Workstation as src

Stage 9: search

| `windows_unusual_count_of_invalid_users_failed_to_auth_using_ntlm_filter`

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
EventCodeeq
  • 4776 corpus 5 (splunk 5)
Statuseq
  • 0xc0000064 corpus 2 (splunk 2)
TargetUserNamene
  • *$ corpus 14 (splunk 14)
isOutliereq
  • 1 corpus 28 (splunk 28)