Security Command Center is Google Cloud's native posture, vulnerability and threat detection service, and most organisations either under-use it or drown in its findings. It brings misconfiguration checks, compliance frameworks, attack path simulation and log-based threat detection into one findings model, but what you actually get depends on the service tier, whether you activate at organisation or project level, and which logs you have enabled. The tier line-up is also changing: the Enterprise tier is deprecated and Google states it will shut down on 21 May 2027, with organisations moved to Premium. This guide explains how to scope, secure and operate Security Command Center (SCC) for practitioners running Google Cloud estates. It complements the Google Cloud Security Guide; tier details here were checked against Google's documentation on 27 October 2026 and change often.
This is an independent guide from NHI Mgmt Group, based on Google Cloud's public documentation as of November 2026. It is not sponsored or reviewed by Google Cloud. Product features change, so confirm settings against Google Cloud's current documentation.
Key takeaways
- Activate at organisation level wherever you can. Project-level activation is available only for Premium and cannot use attack exposure scores or attack paths.
- Know your tier and its future. Standard is free and basic; Premium adds threat detection, attack paths and full Compliance Manager frameworks; Enterprise (Deprecated) moves to Premium on or after 21 May 2027, so plan for losing multi-cloud features that are Enterprise-only.
- Threat detection needs logs. Event Threat Detection reads Cloud Logging; several exfiltration rules need Data Access audit logs that are off by default for most services.
- SCC roles are privileged. Whoever can edit mute rules, notification configs or custom modules can hide an attack. Grant viewer roles by default and edit roles narrowly.
- Service accounts dominate the threat findings. Leaked keys, dormant service accounts, Token Creator grants and anomalous impersonation are core detections; route them to the team that owns NHI response.
How identity and access work in Security Command Center
SCC is activated on an organisation or a project and then reads configuration, asset inventory and logs across that scope. Everything it produces is a finding attached to a resource, with a findingClass, a category, a severity, a state and a mute status. Access to findings follows IAM on the resource hierarchy, so the activation level and role grants together decide who sees what.
Service tiers
| Tier | Scope and activation | Main capabilities | Notes |
|---|---|---|---|
| Standard | Google Cloud only; organisation-level activation | Basic posture management, Compliance Manager with the Security Essentials framework only | No additional charge |
| Standard-legacy | Google Cloud only | Basic posture and compliance | Some organisations may be migrated to Standard |
| Premium | Google Cloud only; project or organisation level | Advanced posture, threat detection (Event, Container and VM Threat Detection), attack path simulation, toxic combinations and chokepoints, full Compliance Manager framework library and audit mode | Pay-as-you-go or subscription; attack paths need organisation activation |
| Enterprise (Deprecated) | Google Cloud, AWS and Azure; organisation level only | Most Premium services plus multi-cloud connectors, CIEM and Google Security Operations SIEM and SOAR integration | Shuts down 21 May 2027; moves to Premium on or after that date |
If you depend on Enterprise for AWS or Azure coverage, Google's tier documentation lists multi-cloud support as an Enterprise feature, so decide now how those clouds will be covered after the migration.
Finding classes
| Class | What it means | Expected response |
|---|---|---|
| Threat | Potential active attack or malicious activity | Investigate immediately; page on high and critical severity |
| Vulnerability | Software flaw that could be exploited | Patch within your vulnerability SLAs |
| Misconfiguration | Insecure setting on a resource or policy | Fix the configuration, ideally in infrastructure as code |
| Posture violation | Drift from your posture or a Compliance Manager cloud control | Remediate or record an approved exception |
| Toxic combination and Chokepoint | Issues that together form an attack path to high-value resources, or where many paths converge | Prioritise above isolated findings |
| Observation | Detail that may matter if the environment is compromised | Review in context |
| SCC error | SCC itself cannot detect issues, for example a missing permission | Fix promptly: these cannot be muted and mean blind spots |
Google also lists External exposure and Secret classes in Preview for Premium.
Admin roles and privileged access
SCC has its own predefined roles under roles/securitycenter.*. Because SCC decides what security teams see, edit rights over it are effectively the right to switch detection off.
| Role | Grants | Suggested holders |
|---|---|---|
roles/securitycenter.admin | Full control, including settings, tier and source IAM policies; with roles/resourcemanager.organizationAdmin it changes organisation configuration | Two or three platform security engineers via just-in-time elevation |
roles/securitycenter.adminEditor | Read-write on findings, mute configs, notification configs, BigQuery exports and custom modules; not tier or organisation settings | Detection engineering team; the pipeline that deploys SCC configuration |
roles/securitycenter.adminViewer | Read access across SCC, including assets, findings and attack paths when granted at organisation level | SOC analysts, auditors |
roles/securitycenter.findingsEditor | View and edit findings on resources in scope | Triage leads who change finding state |
roles/securitycenter.findingsViewer, roles/securitycenter.assetsViewer | Read findings or assets; grant at folder or project level to limit visibility | Application teams for their own projects |
- Least privilege by hierarchy: organisation-level grants inherit everywhere. Give application teams folder or project scoped viewer roles so they see their own findings without seeing every project.
- Treat mute rights as detection rights: a dynamic or static mute rule hides findings from the default view while scanning continues. Review every mute rule like a firewall exception, with an owner and an expiry.
- Protect custom modules: Security Health Analytics and Event Threat Detection custom modules change what is detected; deploy them from version control.
- Elevate, do not standing-grant: manage SCC admin through Google Cloud Privileged Access Manager or your PAM tool. See the Privileged Access Management Guide.
Integrations, API tokens and service accounts
SCC is both a consumer and a producer of machine identities: it relies on service agents to read your estate and send findings, and many of its most valuable detections are about your service accounts.
- Notification service agent: finding notifications use a
NotificationConfigthat publishes to a Pub/Sub topic. Google creates the service agentservice-org-ORGANIZATION_ID@gcp-sa-scc-notification.iam.gserviceaccount.comand grants itsecuritycenter.notificationServiceAgenton the topic. Do not remove that binding during IAM clean-ups, or alerts stop silently. - Inside a perimeter: if the topic project sits inside a VPC Service Controls perimeter, Google notes you may need ingress and egress rules for the caller and an ingress rule for the notification service agent. See the VPC Service Controls Guide.
- Downstream automation: Cloud Run functions or SOAR playbooks triggered from Pub/Sub run as service accounts that can change resources. Scope them to the remediation they perform, not Editor.
- SIEM export: continuous exports to Pub/Sub or BigQuery feed Google Security Operations or another SIEM. See the Google Security Operations (Chronicle) Guide.
Service account detections worth routing to NHI owners
| Event Threat Detection finding | What it signals | ATT&CK |
|---|---|---|
| Initial Access: Leaked Service Account Key Used | A key known to be exposed is in use | T1078.004 Cloud Accounts |
| Initial Access: Dormant Service Account Action | An inactive service account suddenly acts | T1078.004 |
| Persistence: Service Account Key Created | New user-managed key, a durable credential | T1098.001 Additional Cloud Credentials |
| Persistence: IAM Anomalous Grant | Sensitive role granted unusually, for example to an external principal | T1098.003 Additional Cloud Roles |
| Defense Evasion: Organization, Folder or Project level TokenCreatorRole Added | Someone can now mint tokens for service accounts at that scope | T1098.003 |
| Privilege Escalation: Anomalous Impersonation of Service Account for Admin Activity | Unusual impersonation chain used for admin actions | T1078.004 |
| Exfiltration: BigQuery Data Exfiltration | Data copied out of a BigQuery dataset to an outside location | T1537 Transfer Data to Cloud Account |
Google has also added detections for AI agent identities, such as a sensitive role granted by or to an AI agent. See the Google Cloud Service Account Keys Security Guide for prevention of the key-related findings.
Logging and detection
SCC's threat detection is only as good as the telemetry it reads.
- Event Threat Detection sources: Cloud Audit Logs (Admin Activity, Data Access, and service logs for IAM, BigQuery, Cloud SQL, AlloyDB and GKE), Google Workspace Login and Admin audit logs at organisation level, Cloud DNS logs, Cloud IDS logs and SSH auth logs for brute force detection.
- Turn on what the rules need: Admin Activity logs are always on, but the BigQuery and Cloud SQL exfiltration rules need Data Access logs, and some Cloud SQL rules need pgAudit or MySQL database auditing. Enable Data Access logs at least for projects with sensitive data. See the Google Cloud Audit Logs and Log Sinks Guide.
- Runtime detectors: Container Threat Detection and Virtual Machine Threat Detection cover GKE workloads and Compute Engine instances; confirm they are enabled for the projects that matter.
- Sensitive data enrichment: with Sensitive Data Protection discovery, Event Threat Detection can raise the severity of a finding by one level, or change its category, when it affects sensitive data.
- Watch SCC itself: alert on SCC error findings and on changes to mute rules, notification configs and module enablement in Admin Activity logs.
Attack paths and high-value resources
Attack path simulation (Premium or Enterprise, organisation activation) runs on a model of your environment up to every 48 hours and scores resources from 0 to 10. It only works well if you tell it what matters: create resource value configurations (up to 100 per organisation) that mark resources HIGH (10), MEDIUM (5), LOW (1) or NONE (0) by type, scope and tags or labels. A score of 0 means no simulated path was found, not that there is no risk; simulations do not model insiders or zero-days.
Metrics
- Share of projects covered by an organisation-level activation and by threat detection services (target 100 percent of production).
- Mean time to triage high and critical Threat findings, and to remediate critical toxic combinations.
- Count of active mute rules without an owner or expiry, trending to zero.
- Open SCC error findings (target zero).
- Number of active leaked or dormant service account findings older than seven days.
- Compliance Manager pass rate for your chosen framework, tracked per folder.
Hardening baseline
Posture: Compliance Manager and Security Health Analytics
Google is moving posture checks to Compliance Manager. Its model is rules grouped into cloud controls, mapped to regulatory controls and collected into frameworks that you deploy to an organisation, folder or project. The Security Essentials framework is available in all tiers and is the only one on Standard; Premium adds built-in frameworks including CIS Google Cloud Computing Platform Benchmark v3.0, CIS Controls 8.0, NIST SP 800-53 R5, ISO 27001:2022, PCI DSS 4.0 and SOC 2. Controls run in detective mode; audit mode, which gathers evidence, is Premium only. Batch scans run every 16 hours on Premium and every 38 hours on Standard, with near-real-time scans on configuration change in Premium.
Security Health Analytics, the older detector service with categories such as PUBLIC_BUCKET_ACL, OPEN_FIREWALL, OPEN_SSH_PORT, PUBLIC_SQL_INSTANCE and MFA_NOT_ENFORCED, is disabled for new Premium activations according to Google, which recommends Compliance Manager instead. Where both run, they can produce duplicate findings; disable Security Health Analytics where your frameworks cover its detectors, or mute the duplicates. If you used the security posture service, Google recommends deploying a matching framework and then deleting the posture deployment. For the benchmark itself, see the CIS Cloud Foundations Benchmarks Guide.
Rollout steps
- Decide tier and activation: organisation-level Premium for most production estates; Standard as a free baseline for smaller organisations, accepting no managed threat detection. Document how AWS and Azure will be covered if you rely on Enterprise.
- Enable telemetry first: Data Access logs for sensitive projects, Cloud DNS logging, and Workspace log sharing at organisation level.
- Deploy a framework to a pilot folder: start with Security Essentials or the CIS benchmark framework on one folder, review findings with the owning teams, then extend.
- Define high-value resources: create resource value configurations for regulated datasets, key projects and production clusters.
- Wire notifications: one
NotificationConfigper routing need (for examplestate="ACTIVE"Threat findings to the SOC), plus a BigQuery export for reporting. - Tune with expiring mutes: use dynamic mute rules with expiry for accepted risk, never permanent static mutes for whole categories.
- Roll back safely: framework deployments and modules are configuration; keep them in code so a noisy deployment can be removed from a folder and redeployed after tuning.
Common misconfigurations
- Project-level activation by default: chosen for budget, then the organisation loses attack paths, toxic combinations and cross-project visibility.
- Threat detection blind to data access: Data Access logs are off, so BigQuery exfiltration rules never fire despite Premium licensing.
- Mute rules as noise control: broad static mutes on whole categories, created months ago, with no owner. Muted findings still count in compliance percentages, so reports look worse than the queue suggests and real issues stay hidden.
- Broken notification path: an IAM tidy-up removes the notification service agent's role on the topic, or a perimeter blocks it, and alerts stop with no error in the SOC.
- Everyone is an SCC admin: application teams are given
securitycenter.adminat organisation level to see their findings, which also lets them mute and reconfigure detection. - Leaked key findings left open: leaked or dormant service account findings sit unresolved because no team owns machine identities. The Google API Keys Exposure 2026 case shows how quickly exposed Google credentials become a data risk once their privileges change.
Framework mapping
| Framework | Reference | SCC capability |
|---|---|---|
| NIST SP 800-53 Rev. 5 | CA-7 Continuous Monitoring; CM-6 Configuration Settings; RA-5 Vulnerability Monitoring and Scanning; SI-4 System Monitoring | Compliance Manager controls, vulnerability findings, Event Threat Detection |
| NIST CSF 2.0 | ID.RA-01; PR.PS-01; DE.CM-09; DE.AE-02 | Vulnerability identification, configuration management, monitoring and analysis of adverse events |
| CIS Controls v8.1 | 4.1 Secure Configuration Process; 7.5 Automated Vulnerability Scans of Internal Enterprise Assets; 8.11 Conduct Audit Log Reviews | Framework deployments, vulnerability assessment, threat findings review |
| ISO/IEC 27001:2022 Annex A | A.8.8 Management of technical vulnerabilities; A.8.9 Configuration management; A.8.16 Monitoring activities | As above, with audit mode evidence |
| MITRE ATT&CK | T1078.004; T1098.001; T1098.003; T1537 | Service account and exfiltration detections listed above |
Practitioner checklist
- Confirm tier and activation level, and record the Enterprise migration plan if it applies before 21 May 2027.
- Activate at organisation level and confirm threat detection services cover all production projects.
- Enable Data Access audit logs for projects holding sensitive data.
- Deploy a Compliance Manager framework and remove duplicate Security Health Analytics coverage.
- Create resource value configurations for your high-value resources.
- Grant viewer roles by default; restrict admin and adminEditor to a small group and a deployment pipeline.
- Give every mute rule an owner and expiry, and prefer dynamic rules.
- Test the Pub/Sub notification path end to end, including the service agent binding.
- Route service account threat findings to the NHI owner with a response SLA.
- Alert on SCC error findings and on changes to SCC configuration.
Standards and references
- Google Cloud: Security Command Center overview
- Google Cloud: Security Command Center service tiers
- Google Cloud: Event Threat Detection overview
- Google Cloud: Compliance Manager overview
- Google Cloud: Attack exposure scores and attack paths
- Google Cloud: Access control with IAM for Security Command Center
- Google Cloud: Mute findings in Security Command Center
- NIST: SP 800-53 Rev. 5
Related NHI Mgmt Group resources: Google Cloud Security Guide · Google Cloud IAM and Service Accounts Guide · Google Cloud Enterprise Foundations Blueprint Guide · VPC Service Controls Guide