Most organisations have logs switched on somewhere, yet fail CIS Control 8 because they cannot show which assets log, where the logs go, how long they stay and who looked at them last week. CIS Control 8 (Audit Log Management) in CIS Controls v8.1 sets twelve Safeguards that take an organisation from a written logging process to centralised, retained and regularly reviewed logs, including DNS, URL, command-line and service provider logs. This guide explains each Safeguard by Implementation Group, the identity and non-human identity (NHI) logging that assessors increasingly ask about, an implementation sequence, the evidence and measures in the CIS Controls Assessment Specification, and how the Control maps to other frameworks. It is for security operations leads, infrastructure owners and GRC staff. It does not repeat general log management practice such as storage tiering or retention by regulation: for that, see the Log Management and Retention Guide.
Key takeaways
- IG1 is a process plus logging that actually works. Safeguards 8.1 to 8.3 require a documented audit log management process, logging enabled across enterprise assets and enough storage at each logging destination that logs are not overwritten.
- IG2 is where Control 8 becomes a detection capability. Time synchronisation, detailed logs for sensitive assets, DNS, URL and command-line logging, centralisation, 90-day retention and weekly review (8.4 to 8.11) all sit at IG2.
- Two numbers fail assessments outright. The assessment specification fails 8.1 if the process has not been reviewed in more than twelve months, and fails 8.11 if more than seven days pass between log reviews.
- 90 days is the CIS floor, not a recommendation. Safeguard 8.10 sets a minimum of 90 days; payment card, government and sector rules often require a year or more, so take the longest requirement that applies.
- Machine identities must appear in the logs. Service principal, managed identity and API key activity is often missing from collection, and command-line logs can capture secrets, so Control 8 needs an explicit NHI scope.
What it is and who it applies to
CIS summarises Control 8 as collecting, alerting on, reviewing and retaining audit logs of events that could help detect, understand or recover from an attack. It applies to every enterprise asset that can produce logs, from servers and endpoints to network devices, SaaS platforms and cloud control planes.
- IG1: CIS calls IG1 essential cyber hygiene. For Control 8 that is 8.1 to 8.3. A small organisation can meet it by writing a one or two page logging process, turning on native logging in its operating systems, Microsoft 365 or Google Workspace tenant and firewall, and checking that local log sizes and rotation do not discard events within the period the process states.
- IG2: adds 8.4 to 8.11. These need a central log platform (a SIEM, a log management service or a cloud-native log store) and someone who reviews it at least weekly. This is the realistic target for organisations with dedicated IT or security staff. See the SIEM Guide.
- IG3: adds 8.12, collection of service provider logs. In practice many IG2 organisations already need SaaS and cloud provider logs, because that is where identity and data activity now happens.
CIS Controls v8.1 (June 2024) clarified Safeguard wording, added a Governance security function and new asset classes, and realigned mappings to NIST CSF 2.0. CIS states that no Implementation Group assignments changed. The Safeguard numbers and titles below are unchanged from v8; the asset type and security function values quoted come from the v8 assessment specification, so take v8.1 values from the CIS v8.1 spreadsheet if you report them.
Structure and key requirements
| Safeguard | Title | IG | What it requires |
|---|---|---|---|
| 8.1 | Establish and Maintain an Audit Log Management Process | IG1, IG2, IG3 | A process defining logging requirements; at minimum collection, review and retention; reviewed annually or on significant change |
| 8.2 | Collect Audit Logs | IG1, IG2, IG3 | Logging enabled across enterprise assets as the 8.1 process defines |
| 8.3 | Ensure Adequate Audit Log Storage | IG1, IG2, IG3 | Logging destinations keep enough storage to meet the process (rotation, maximum size) |
| 8.4 | Standardize Time Synchronization | IG2, IG3 | At least two synchronised time sources configured across assets, where supported |
| 8.5 | Collect Detailed Audit Logs | IG2, IG3 | Detailed logs on assets holding sensitive data: event source, date, username, timestamp, source and destination addresses and other forensic elements |
| 8.6 | Collect DNS Query Audit Logs | IG2, IG3 | DNS query logs, where appropriate and supported |
| 8.7 | Collect URL Request Audit Logs | IG2, IG3 | URL request logs, where appropriate and supported |
| 8.8 | Collect Command-Line Audit Logs | IG2, IG3 | Command-line logs, for example PowerShell, BASH and remote administrative terminals |
| 8.9 | Centralize Audit Logs | IG2, IG3 | Central collection and retention to the extent possible |
| 8.10 | Retain Audit Logs | IG2, IG3 | Retention for a minimum of 90 days |
| 8.11 | Conduct Audit Log Reviews | IG2, IG3 | Reviews to detect anomalies, weekly or more often |
| 8.12 | Collect Service Provider Logs | IG3 | Service provider logs where supported, for example authentication and authorisation, data creation and disposal, and user management events |
Turning the Safeguards into concrete settings
| Safeguard | Typical source | Concrete setting or event |
|---|---|---|
| 8.2, 8.5 | Windows | Advanced Audit Policy subcategories (logon, account management, policy change); set "Audit: Force audit policy subcategory settings (Windows Vista or later) to override audit policy category settings" so basic policy cannot overwrite them (event 4719 records such a change). See the Windows Event Log Guide. |
| 8.8 | Windows process creation | Audit Process Creation (Detailed Tracking) plus "Include command line in process creation events" (Administrative Templates, System, Audit Process Creation); both are needed for command lines in event 4688 |
| 8.8 | PowerShell | "Turn on PowerShell Script Block Logging" writes event 4104 to the PowerShell Operational log; add Protected Event Logging so sensitive script content is encrypted. See the PowerShell Security Guide. |
| 8.8 | Linux | auditd rules for process execution (execve) and privileged commands, forwarded with the host's syslog agent. See the Linux Logging and auditd for Defenders Guide. |
| 8.6 | Internal DNS resolvers, protective DNS | Query logging on resolvers; the assessment specification measures configured versus total internal DNS servers |
| 8.7 | Web proxy, secure web gateway, browser telemetry | URL, user, device and verdict per request; see the Secure Web Gateway and Web Filtering Guide |
| 8.4 | All logging assets | Two approved NTP sources per asset, alerting on drift; see the NTP Security Guide |
| 8.12 | Identity provider, SaaS, cloud control plane | Sign-in and audit logs exported by API or native connector before platform retention expires; see the Cloud Logging and Audit Trail Guide |
"Where appropriate and supported" in 8.6 and 8.7 is not an exemption for everything. Document in the 8.1 process which assets cannot produce DNS or URL logs and where the equivalent visibility comes from instead (resolver or proxy rather than endpoint), so the assessor sees a decision rather than a gap.
Identity and NHI requirements
Control 8 is the Control most organisations rely on to answer "who did this?", and the answer is increasingly a machine identity. Four requirements belong in the 8.1 process.
- Collect every sign-in category: Microsoft Entra ID separates sign-in logs into interactive user, non-interactive user, service principal and managed identity sign-ins. Collecting only interactive sign-ins, a common default in older integrations, leaves most automated activity invisible. Apply the same check to other identity providers and cloud platforms. See the Cloud Logging and Audit Trail Guide.
- Make 8.5 "username" mean an owned identity: detailed logs for sensitive assets should record which service account, API key identifier or workload identity acted, not just a shared technical account. Shared accounts defeat the forensic purpose of 8.5. See the Service Account Security Guide.
- Treat command-line logs as sensitive data: Microsoft warns that enabling command lines in event 4688 records arguments in plain text, readable by anyone with rights to read the Security log, and command lines can contain passwords and tokens. Restrict log read access, mask secrets in the pipeline where possible, and treat any secret found in logs as exposed and due for rotation. See the Secrets Management Guide.
- Protect the logging pipeline's own identities: forwarders, SIEM connectors and cloud export roles hold read access to many systems and write access to the log store. An attacker who controls them can stop collection quietly. Scope them tightly, vault their credentials and alert when they change. After the Sumo Logic breach, in which a compromised credential was used to access one of the log platform's AWS accounts, customers were asked to rotate API access keys and collector credentials used with the service: keep an inventory of these credentials so that rotation is possible in hours.
Licence tier matters for 8.12. In the Storm-0558 key breach, the US State Department detected forged-token mailbox access through a custom alert on the MailItemsAccessed audit event, which at the time came with premium logging; organisations without it could not see the same activity, and Microsoft later extended that logging to all customers. Check that the events your 8.1 process needs are actually available in your service plan.
Implementation steps
- Write the process (8.1): list log sources by category (identity, endpoint, server, network, cloud, SaaS, security tools), what each must log, the destination, retention, who reviews and how often, and the exceptions where a source cannot log. Assign an owner and review it annually. See the Logging and Monitoring Policy Template.
- Enable baseline logging (8.2) through configuration standards: put audit settings in the Control 4 secure configuration baselines (Group Policy, Intune, configuration management) rather than enabling them by hand, so new assets inherit them. Start from the Control 1 asset inventory to find assets capable of logging.
- Size storage (8.3): set local log sizes and rotation so events survive until they are forwarded, and decide what happens when a destination fills (overwrite, stop or alert). NIST SP 800-53 AU-5 covers response to logging failures.
- Synchronise time (8.4): configure two approved internal or trusted NTP sources on every asset that supports it and check offsets as part of the log health dashboard. Correlation across sources fails without consistent timestamps.
- Centralise in waves (8.9): pilot forwarding from identity, domain controllers and internet-facing systems first, measure ingestion and cost, then add servers, endpoints, DNS (8.6), proxy (8.7) and command-line logs (8.8). Keep local logging on during the move so a forwarder failure does not create a gap; the rollback path is to revert to local collection, never to disable logging.
- Add detail on sensitive assets (8.5): use the Control 3 data inventory to identify assets storing sensitive data and enable data access and detailed audit logging there.
- Set retention (8.10): configure the central store for at least 90 days searchable, longer where regulation or contract requires, and protect it from deletion with immutable or write-once storage held in a separate account.
- Bring in service providers (8.12): from the Control 15 service provider inventory, list providers that expose logs, connect them and record those that do not as a supplier risk.
- Start weekly review (8.11): define what a review covers (alerts closed, sources silent, failed log-ons, privileged changes, NHI anomalies), who performs it and where the record is kept. Automate detections so the human review is of exceptions, not raw logs.
Small versus large organisations
- Small (IG1, early IG2): use the logging built into your productivity suite, endpoint protection and firewall, export to a single managed log store, and use a managed detection service for weekly review if you lack staff. Confirm the retention your licence actually gives.
- Large (IG2, IG3): the problems are coverage drift, cost and ownership. Track silent sources automatically, filter low-value events in the pipeline rather than at the source, and give each log source a named owner. See the Open Source SIEM and Detection Tools Landscape Guide for self-managed options.
Evidence and assessment
The CIS Controls Assessment Specification measures Control 8 mainly through counts of assets capable of logging and of those configured correctly, plus dates.
| Safeguard | Evidence an assessor expects | Measure to report |
|---|---|---|
| 8.1 | Approved process covering collection, review and retention, with version history | Months since last review (fails above 12) |
| 8.2 | Configuration baselines showing audit settings; list of assets capable of logging | Assets correctly configured to log versus assets capable of logging |
| 8.3 | Log size and rotation settings per asset type; destination capacity reports | Assets configured to meet retention requirements |
| 8.4 | List of approved time sources; configuration export | Assets using at least two approved sources; assets using unapproved or no sources |
| 8.5 | Sensitive asset list (from Control 3) and their detailed logging configuration | Sensitive assets correctly configured for detailed logging |
| 8.6, 8.7, 8.8 | Resolver, proxy and endpoint audit configurations; sample events | Capable assets configured versus not configured |
| 8.9 | Log aggregation software in the software inventory; source coverage report | Logging assets covered by at least one aggregator |
| 8.10 | Retention settings on each aggregator | Aggregators retaining 90 days or more |
| 8.11 | Dated review records (tickets, reports, sign-offs) at regular intervals | Days between consecutive reviews (fails above 7) |
| 8.12 | Service provider inventory marked with log support and connection status | Providers configured for log collection versus providers that support it |
Metrics that show it is working
- Source coverage: percentage of logging-capable assets (Control 1 inventory) sending to the central platform, reported by asset class.
- Silent sources: count of sources with no events in the expected window; a silent domain controller or firewall is an incident until proven otherwise.
- NHI coverage: percentage of identity providers and cloud accounts from which service principal, workload identity and API key activity is collected.
- Retention compliance: oldest searchable event per source against the 8.1 requirement, plus a tested restore time for archived logs.
- Review cadence: days between documented reviews and the share of review findings turned into tickets or detection changes.
Detect attacks on the logs themselves: alert on T1685.005 Clear Windows Event Logs (event 1102 in the Security log), T1685.001 Disable or Modify Windows Event Log (audit policy changes such as event 4719, stopped event log service) and on gaps in forwarding. Command-line and script block logging (8.8) are the main data sources for T1059.001 PowerShell and T1059.004 Unix Shell, and DNS logs (8.6) for T1071.004 DNS command and control.
Mapping to other frameworks
| Safeguards | NIST CSF 2.0 | NIST SP 800-53 Rev 5 | ISO/IEC 27001:2022 Annex A |
|---|---|---|---|
| 8.1 | PR.PS-04 | AU-1, AU-2 | A.8.15 Logging |
| 8.2, 8.5 | PR.PS-04 | AU-2, AU-3, AU-12 | A.8.15 |
| 8.3, 8.10 | PR.PS-04 | AU-4, AU-5, AU-11 | A.8.15 |
| 8.4 | PR.PS-04 | AU-8 | A.8.17 Clock synchronisation |
| 8.6, 8.7, 8.8 | DE.CM-01, DE.CM-09 | AU-12, SI-4 | A.8.15, A.8.16 Monitoring activities |
| 8.9 | DE.AE-03 | AU-6(3), AU-9(2) | A.8.15 |
| 8.11 | DE.AE-02, DE.CM-03 | AU-6 | A.8.16 |
| 8.12 | DE.CM-09 | AU-2, SI-4 | A.8.15, A.5.22 Monitoring, review and change management of supplier services |
This crosswalk is a practitioner mapping built from control text; CIS publishes official v8.1 mappings, and those should be used for formal attestations. NIST SP 800-92 Rev. 1 (initial public draft, October 2023), the Cybersecurity Log Management Planning Guide, is a useful companion for writing the 8.1 process. The joint event logging best practice guidance led by the Australian Signals Directorate with CISA, NCSC-UK and other partners sets a baseline that aligns well with IG2. See the NIST SP 800-92 Log Management Guide.
Common pitfalls
- Logging enabled, never centralised: events sit on endpoints and rotate away before anyone needs them. 8.2 evidence looks fine while 8.9 and 8.10 fail; measure coverage at the central store, not at the source.
- Logs the attacker can delete: the central store sits in the same administrative domain as the systems it logs, so a domain or cloud admin compromise removes the evidence. Keep log storage and its admin roles separate and immutable.
- Silent failures: a forwarder breaks, a SaaS connector token expires or a licence lapses, and nobody notices for weeks. Alert on missing data per source, and treat expired connector credentials as an NHI lifecycle issue.
- Licence-tier blind spots: the events needed to investigate mailbox or data access are not in the plan the organisation pays for, as the Storm-0558 case showed. Map required events to the licence during 8.1 review.
- Human sign-ins only: service principal, managed identity and API key activity is excluded from collection, so a compromised automation credential leaves no trace. Review sign-in and audit coverage for every non-human identity type the platform supports.
- Weekly review as a ritual: someone ticks a box without a defined scope. Write down what the review checks, record findings and feed them into detection rules, which is what keeps 8.11 useful as well as compliant.
Practitioner checklist
- Confirm your target Implementation Group: IG1 needs 8.1 to 8.3, IG2 adds 8.4 to 8.11, IG3 adds 8.12.
- Approve an audit log management process covering sources, content, destination, retention, review and exceptions; review it at least every twelve months.
- Put audit settings in configuration baselines, including Advanced Audit Policy enforcement, event 4688 command lines and PowerShell script block logging.
- Configure at least two approved time sources on every logging asset and monitor drift.
- Centralise logs from identity, servers, endpoints, DNS, proxy, cloud and SaaS, keeping local logging during migration.
- Collect service principal, managed identity and API key activity, and restrict read access to logs that can contain secrets.
- Retain at least 90 days searchable, longer where rules require, in immutable storage under separate administration.
- Run and record a scoped log review at least weekly, and alert on silent sources and log clearing.
- Connect service providers that expose logs and record those that do not as supplier risks.
Standards and references
- CIS: Control 8 Audit Log Management
- CIS: Controls Assessment Specification, Control 8
- CIS: The Ongoing Evolution of the CIS Critical Security Controls
- NIST: SP 800-92 Rev. 1 Cybersecurity Log Management Planning Guide (initial public draft)
- CISA and partners: Best Practices for Event Logging and Threat Detection
- NCSC: Logging and protective monitoring
- Microsoft: Command line process auditing
- Microsoft: about_Logging_Windows
Related NHI Mgmt Group resources: CIS Controls Guide · Log Management and Retention Guide · SIEM Guide · CIS Control 7 Continuous Vulnerability Management Guide · CIS Control 9 Email and Web Browser Protections Guide