Detection rules › Splunk

Windows Unusual Count Of Users Failed To Authenticate From Process

Author
Mauricio Velazco, Splunk
Source
upstream

The following analytic identifies a source process failing to authenticate multiple users, potentially indicating a Password Spraying attack. It leverages Windows Event 4625, which logs failed logon attempts, and uses statistical analysis to detect anomalies. This activity is significant as it may represent an adversary attempting to gain initial access or elevate privileges within an Active Directory environment. If confirmed malicious, the attacker could compromise multiple accounts, leading to unauthorized access, data exfiltration, or further lateral movement within the network.

MITRE ATT&CK coverage

TacticTechniques
Credential AccessT1110.003 Brute Force: Password Spraying

Event coverage

ProviderEvent IDTitle
Security-Auditing4625An account failed to log on.

Stages and Predicates

Stage 1: search

search EventCode=4625 Logon_Type=2 ProcessName!="-"

Stage 2: bucket

bucket span=2m _time

Stage 3: stats

stats dc(TargetUserName) AS unique_accounts,AS user,AS dest,AS src BY _time, ProcessName, SubjectUserName, Computer, action, app, authentication_method, signature, signature_id

Stage 4: eventstats

eventstats avg(unique_accounts) AS comp_avg BY ProcessName, SubjectUserName, Computer

Stage 5: eval

eval ... using (comp_avg, comp_std)

Stage 6: eval

eval ... using (unique_accounts, upperBound)

Stage 7: search

search isOutlier=1

Stage 8: search

search `macro`

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
  • 4625 corpus 6 (splunk 6)
Logon_Typeeq
  • 2 corpus 2 (splunk 2)
ProcessNamene
  • "-" corpus 2 (splunk 2)
isOutliereq
  • 1 corpus 16 (splunk 16)

Neighbors

Broader alternatives (more inclusive than this rule)

These rules match a superset of what this rule catches. They cover the same events plus more. Use them if you want wider coverage and can absorb more false positives.