Join our Newsletter — 33% off our NHI Course
Home› Guides› MITRE ATT&CK T1685.001 Disable or Modify Tools: Detecting…
Guide SOC & Detection Engineering

MITRE ATT&CK T1685.001 Disable or Modify Tools: Detecting Windows Event Log and Audit Policy Tampering

PDF for newsletter subscribers · free to read online

← All guides
By NHI Mgmt Group, edited by Lalit Choda Updated 14 November 2026 14 min read
On this page

If an attacker turns off Windows auditing before acting, every detection that depends on the Security log goes quiet, and the SOC sees an absence rather than an alert. MITRE ATT&CK T1685.001 Disable or Modify Windows Event Log is the sub-technique of T1685 Disable or Modify Tools, in the Defense Impairment tactic (TA0112), that covers stopping or disabling the EventLog service, changing audit policy, disabling individual log channels and editing the Autologger sessions that feed them. This guide goes deeper than the T1685 Disable or Modify Tools Defensive Guide: it covers the specific event IDs, registry locations and tool usage that reveal event log tampering, sample detection logic, the controls that make tampering hard, and how to respond. Clearing logs is a separate sub-technique, T1685.005 Clear Windows Event Logs, and is only touched on here. It is for detection engineers, Windows platform owners and incident responders, and covers defence only.

Key takeaways

  • Tampering has a small, well-defined footprint. Security 4719 (audit policy changed), Security 1100 (EventLog service shut down), System 7035 (service control), Sysmon 13 on EventLog, WINEVT channel and Autologger keys, and process creation for auditpol and wevtutil cover most methods.
  • 4719 is always written. Microsoft documents that the audit policy change event is logged regardless of the Audit Policy Change subcategory setting, so treat every unplanned 4719 as an investigation.
  • The last event before silence matters most. Forward events off the host in near real time, because once logging stops locally the only copy of the tampering evidence is the one already collected.
  • Detect the gap as well as the change. Some methods take effect only after a reboot or leave no event; per-host heartbeat and per-channel volume analytics catch what event rules miss.
  • It needs administrator rights, mostly. ATT&CK notes one Autologger route works without admin rights before a reboot; the rest require local admin, so privilege reduction is the main preventive control.

What T1685.001 covers

ATT&CK v20 describes the Windows Event Log as the record of user and system activity, such as logons and process creation, that security tools and analysts rely on, and the EventLog service as the component that maintains it. The audit policy, set through Local Security Policy, Group Policy or auditpol, decides which security events are written. The sub-technique groups four kinds of tampering:

MethodWhat changesEffectPrimary evidence
EventLog service stopped or disabledService state or the Start value under HKLM\SYSTEM\CurrentControlSet\Services\EventLogAll channels stop recording, immediately or after rebootSecurity 1100, System 7035, Sysmon 13 on the service key, then a gap
Audit policy reduced or clearedBasic or advanced audit policy categories and subcategoriesSelected security events, such as logon or process creation, stop being writtenSecurity 4719 with the subcategory and the change, auditpol process creation
Individual channel disabledThe Enabled value under HKLM\SOFTWARE\Microsoft\Windows\CurrentVersion\WINEVT\Channels\<channel>, or wevtutil set-log with the enabled option set to falseOne channel, for example PowerShell Operational or Sysmon, goes silentSysmon 13 with value 0, wevtutil process creation, channel volume drop
Autologger session alteredStart values under HKLM\SYSTEM\CurrentControlSet\Control\WMI\Autologger\EventLog-Security, EventLog-System or EventLog-ApplicationThe affected log stops after the next rebootSysmon 13 on the Autologger keys, then a gap after restart

ATT&CK also lists the MiniNT registry key, which disables Event Viewer, under its registry permissions mitigation, and records Threat Group-3390 using appcmd.exe to disable IIS logging on a server, which shows that application logs are in scope as well as the core channels. Related sub-techniques: T1685.002 for cloud logs, T1685.004 for the Linux audit system and T1685.005 for clearing. For deleting artefacts more broadly, see the T1070 Indicator Removal Defensive Guide.

Who uses it and when

ATT&CK v20 records the following procedures for T1685.001:

  • APT29, SolarWinds compromise: used auditpol to prevent the collection of audit logs.
  • Sandworm Team, 2016 Ukraine electric power attack: disabled event logging on compromised systems.
  • Magic Hound: executed scripts to disable the event log service.
  • Threat Group-3390: used appcmd.exe to disable logging on a victim server.
  • HomeLand Justice campaign: deleted Windows events and application logs.

The pattern is consistent: tampering happens after privileged access is gained and shortly before the actions the attacker most wants hidden, such as credential theft, lateral movement, data theft or ransomware deployment. A 4719 or channel disable on a server is therefore an early warning of a larger intrusion, not an isolated configuration error.

Event ID and telemetry reference

LogEventMeaningDetection use
Security4719System audit policy was changed; fields include Subcategory, Subcategory GUID and Changes (success or failure added or removed)Alert on any "removed" change outside approved Group Policy refreshes
Security1100The event logging service has shut down; also written at normal shutdown, not at emergency resetCorrelate with absence of a matching system restart (System 6006 and 6005)
Security1102The audit log was cleared (T1685.005)Often paired with T1685.001 in the same session; include in the same alert bundle
System7035Service Control Manager sent a control (start or stop) to a serviceATT&CK lists it for EventLog service control in DET0187
System104A non-Security log was clearedCollected in the Microsoft WEF baseline alongside 1102
Sysmon13, 14Registry value set; key or value renamedEventLog service Start, WINEVT Channels Enabled, Autologger Start, MiniNT key
Sysmon / Security1 / 4688Process creation with command lineauditpol, wevtutil, sc, PowerShell service cmdlets, appcmd, reg
Sysmon16Sysmon configuration changed; cannot be filteredSysmon's own channel being reconfigured is the same intent against a different log

The Windows Event Log Guide covers channel architecture and audit policy design; this page assumes it.

Detection patterns and sample logic

ATT&CK's DET0187 (analytic AN0535) combines registry changes, service state changes, process execution and anomalies in event record sequences, with three tuning elements: authorised admin accounts, a correlation time window and the list of monitored services. The pseudo-queries below use Sysmon and Security field names; port them to your SIEM. The SigmaHQ rules "Audit Policy Tampering Via Auditpol" and "Disable Windows Event Logging Via Registry" implement patterns 2 and 3 and are tagged T1685.001.

PatternSample logic (pseudo-query)SeverityTuning notes
1. Audit subcategory removedEventID = 4719 AND Changes contains "removed" AND SubjectUserName NOT IN authorised_admins AND NOT during GPO change windowHighPolicy applied by Group Policy appears with the computer account as subject; compare against the intended policy rather than suppressing the account
2. auditpol tamperingprocess OriginalFileName = AUDITPOL.EXE AND command line contains any of (disable, clear, remove, restore)HighBackup and get are benign; restore is included because it can load a weaker policy
3. Channel disabled via registrySysmon 13 AND TargetObject contains \WINEVT\Channels\ AND ends with \Enabled AND Details = DWORD 0 AND Image NOT IN (wevtutil during installs, DrvInst, TiWorker, TrustedInstaller)HighInstallers register disabled channels; SigmaHQ filters these
4. Channel disabled via wevtutilprocess = wevtutil AND command line contains (sl or set-log) AND enabled option set to falseHighRare in normal operations
5. EventLog service start type changedSysmon 13 AND TargetObject = \Services\EventLog\Start AND Details != DWORD 2CriticalAutomatic start (2) is the expected value
6. Autologger session editedSysmon 13 AND TargetObject contains \WMI\Autologger\EventLog-Security OR EventLog-System OR EventLog-Application AND value name = StartCriticalTakes effect at next reboot; alert immediately so the host can be checked before restart
7. Service stop without restartSecurity 1100 OR System 7035 targeting EventLog, with no System 6006 within 2 minutesCriticalPlanned shutdowns write both; patch windows can be suppressed by maintenance calendar
8. MiniNT key createdSysmon 12 or 13 AND TargetObject = HKLM\SYSTEM\CurrentControlSet\Control\MiniNTHighShould not exist on a running production system
9. Host or channel silencehost last event older than (normal interval x 3) OR Security channel event rate < 10 percent of 14-day median for that hostMedium, raise if pairedExclude hosts known to be powered off; reconcile with asset inventory
10. Record ID discontinuityfor each host and channel, EventRecordID jumps or resets without a 1102 or 104MediumRequires the SIEM to keep record IDs; useful forensic signal

Correlations that raise confidence

  • Tamper then act: any of patterns 1 to 8 followed within 30 minutes by a new service install, LSASS access, remote logon to another host or mass file changes. This is the sequence seen before ransomware deployment.
  • Tamper plus clear: a pattern from this page and a 1102 or 104 in the same logon session (link by SubjectLogonId). Together they indicate deliberate anti-forensics.
  • Fleet-wide change: the same 4719 or registry change on many hosts within minutes, without a matching Group Policy change record, suggests a compromised deployment or management account.

Prevention and hardening

ControlSetting or approachATT&CK mitigation
Least privilegeRemove standing local admin rights; use tiered admin accounts and just-in-time elevation; restrict who can edit audit-related Group Policy objectsM1018 User Account Management
Enforced audit policyDeliver advanced audit policy through domain Group Policy so local changes are reverted at the next refresh, and review auditpol output for admin accounts periodicallyM1047 Audit
Registry permissionsKeep default ACLs on the EventLog, WINEVT Channels and Autologger keys; monitor any ACL change on themM1024 Restrict Registry Permissions
File permissionsProtect .evtx files under C:\Windows\System32\winevt\Logs with default permissionsM1022 Restrict File and Directory Permissions
Off-host collectionWindows Event Forwarding (source-initiated subscriptions to a collector) or an agent forwarding to the SIEM in near real timeSupports detection after local loss (NIST SP 800-53 AU-9(2))
Log sizingSize Security and key Operational channels so events survive collector outages; Microsoft notes the client log is WEF's only buffer and overwrites give no notice of lossSupports AU-5 response to logging failures
Endpoint tamper protectionTurn on EDR and antivirus tamper protection so the agents that watch the event log cannot themselves be switched offSee the parent T1685 guide

Microsoft's WEF guidance is clear that forwarding is passive: it cannot enable channels, change log sizes or set audit policy, so the source device must already be configured correctly. Group Policy enforcement and off-host collection together are what make T1685.001 visible.

Rolling out the detections

  1. Confirm the foundations: advanced audit policy delivered by Group Policy with Audit Policy Change enabled for success and failure; Sysmon or EDR registry monitoring covering the four key paths; command line capture for process creation.
  2. Forward the evidence: add 4719, 1100, 1102, 104, 7035 and the Sysmon registry events for these keys to the baseline forwarding subscription for every Windows host, not only servers.
  3. Turn on high-fidelity rules first: patterns 2, 4, 5, 6 and 8 rarely fire legitimately. Run them in alert mode from day one.
  4. Tune 4719 and channel rules: run patterns 1 and 3 for two weeks as reports, document legitimate sources (GPO refresh, installers), then alert.
  5. Add silence analytics: build per-host heartbeat and per-channel volume baselines (patterns 9 and 10) once the SIEM holds 14 days of data.
  6. Test and roll back safely: exercise each rule on an isolated test host with approved test tooling; if a rule floods the SOC, lower it to a report and fix the filter rather than deleting it.

Small and large estates: small organisations can rely on EDR registry and process rules plus a simple "host stopped reporting" alert, which together cover most methods. Large estates with thousands of servers should invest in the fleet-wide correlation and record ID analytics, and should treat their Group Policy and management tooling accounts as tier 0, since one compromised account can change audit policy everywhere at once.

Responding to event log tampering

  1. Treat it as an intrusion: unplanned audit or logging changes on a production host are presumed hostile until the change owner confirms otherwise.
  2. Find the account: take SubjectUserName and SubjectLogonId from 4719 or the user from the process event, and trace that logon (4624) back to its source host.
  3. Bound the blind period: use the last forwarded event and the first event after recovery to set the window, then fill it from other sources: EDR telemetry, network flow, firewall, proxy and identity provider logs.
  4. Restore logging: force a Group Policy refresh, reset the registry values to the baseline and restart the EventLog service where needed, and confirm events resume at the collector.
  5. Preserve evidence: keep the forwarded events, a copy of the .evtx files, the registry hives and a memory capture if the host has not been rebooted.
  6. Scope the account: if the account is a domain, deployment or service account, check every host it touched in the period for the same changes.

Breach notification decisions often depend on proving what was accessed. A blind period you cannot fill may force a worst-case assumption, so record the gap and the alternative sources used in the incident report. See the If your SIEM is compromised playbook for the case where the collection tier itself is targeted.

Where identity fits

Every method on this page except one Autologger route needs administrator rights, and the fleet-wide variant needs a management or Group Policy identity. The accounts that most often enable T1685.001 at scale are deployment service accounts, configuration management agents and Group Policy delegates, which are frequently non-human identities with broad, standing rights. Inventory them, remove rights they do not use, put them behind privileged access controls and alert when they change audit settings outside their normal schedule. See the Detection Engineering for Non-Human Identities Guide.

Metrics

  • Forwarding coverage: percentage of Windows hosts that sent Security events to the collector in the last 24 hours. Target above 98 percent for servers.
  • Audit policy drift: number of hosts whose effective audit policy differs from the Group Policy baseline, measured weekly.
  • Unplanned 4719 rate: count of 4719 removals without a change record per month, and the percentage investigated.
  • Time to detect silence: median minutes between the last event from a host and the heartbeat alert.
  • Detection test pass rate: percentage of quarterly T1685.001 tests that raised the expected alert.

Common failure patterns

  • Collecting only on servers: workstations are where intrusions often start, and tampering there goes unseen if they do not forward.
  • Batch forwarding: hourly or daily log shipping means the tamper event may never leave the host.
  • Suppressing 4719 as noise: filtering it by account rather than by expected change removes the most direct signal.
  • No silence detection: rules that wait for an event cannot fire when the event source has stopped.
  • Ignoring application and Sysmon channels: disabling PowerShell Operational or Sysmon is the same intent as stopping the Security log, but many rule sets watch only the core logs.
  • Restoring without scoping: fixing the setting on one host without checking the account that made the change leaves other hosts blind.

Framework mapping

FrameworkReferenceRelevance
MITRE ATT&CK v20T1685.001; DET0187; mitigations M1018, M1047, M1024, M1022Technique, analytic AN0535 and mitigations
MITRE ATT&CK v20T1685.005 Clear Windows Event Logs, T1112 Modify Registry, T1070 Indicator RemovalAdjacent techniques in the same sessions
NIST CSF 2.0PR.PS-04, DE.CM-09, DE.AE-02Log records generated and available; monitoring; analysing adverse events
NIST SP 800-53 Rev. 5AU-5, AU-6, AU-9, AU-9(2), AU-12, AC-6Logging failure response, review, protection and off-system storage of audit records, generation, least privilege
ISO/IEC 27001:2022 Annex AA.8.15 Logging, A.8.16 Monitoring activitiesLogs must be produced, protected and analysed

Practitioner checklist

  • Deliver advanced audit policy by Group Policy, including Audit Policy Change for success and failure.
  • Forward 4719, 1100, 1102, 104 and 7035 from every Windows host in near real time.
  • Monitor registry writes to the EventLog service, WINEVT Channels, Autologger and MiniNT paths.
  • Alert on auditpol disable, clear, remove and restore, and on wevtutil set-log disabling a channel.
  • Build heartbeat and per-channel volume analytics for silence.
  • Remove standing local admin and review the accounts allowed to edit audit Group Policy.
  • Size local logs to survive collector outages.
  • Write the response steps, including blind period reconstruction, into the SOC runbook.
  • Test each detection quarterly on an isolated host.

Standards and references

Related NHI Mgmt Group resources: T1685 Disable or Modify Tools Defensive Guide · T1070 Indicator Removal Defensive Guide · Windows Event Log Guide · Sysmon Guide · CIS Control 8 Audit Log Management Guide

PDF for subscribers

Take this guide with you

Every section, table and checklist in one PDF, with links to the sources and related guides. Subscribe to our weekly NHI & AI Identity Journal and we’ll email you a download link.

Free weekly newsletter

Subscribe to the NHI & AI Identity Journal

The latest on NHI and Agentic AI security – articles, research, breaches, news and events every week.

Bonus 33% off our NHI Course when you subscribe.

NHIMG Editorial Note
Researched and written by the NHI Mgmt Group editorial team using AI tools, under the editorial direction of Lalit Choda. Facts are drawn from the sources listed on this page and checked by our automated quality process. Last updated 14 November 2026. Spotted an error? Use our contact page.
NHI Mgmt Group, the independent authority on Non-Human Identity, IAM and Agentic AI security. nhimg.org

PDF Downloads for Newsletter Subscribers

Get the guide

Enter your details and we’ll email you a download link.

The full guide is free to read online. The PDF is a thank you for subscribing.

Fill in all fields marked * and tick the PDF box to continue.