Google Cloud will make a new English version of the Professional Security Operations Engineer exam live on November 24, 2026. Exam guide 2.0 keeps the certification name and the published exam format, but it materially reshapes preparation: data management rises from 14% to 20%, Google Threat Intelligence becomes a first-class platform, and explicit objectives now cover artificial intelligence, Model Context Protocol (MCP), non-human identities, and the Triage and Investigation Agent.
The date matters for anyone already planning an appointment. Google directs candidates taking the English exam on or after November 24 to the new guide. The prior guide is therefore the right baseline for identifying study gaps, without assuming that every language version changes on the same day.
What changes on November 24
The official certification page (opens in a new tab) confirms the November 24 launch and says the update reflects new capabilities in Google Security Operations (SecOps), Security Command Center (SCC), and Google Threat Intelligence (GTI). Google presents this as an updated version of the existing Professional Security Operations Engineer exam, not a new credential or exam code.
| Domain | Previous guide | Guide 2.0 | Change |
|---|---|---|---|
| Platform operations | ~14% | ~13% | slight decrease |
| Data management | ~14% | ~20% | 6-point increase |
| Threat hunting | ~19% | ~18% | slight decrease |
| Detection engineering | ~22% | ~20% | 2-point decrease |
| Incident response | ~21% | ~18% | 3-point decrease |
| Observability | ~10% | ~11% | slight increase |
Those percentages do not make detection and incident response secondary. Together, they still represent about 38% of the exam. The larger data-management domain signals that ingestion, parsing, normalization, enrichment, and log costs can no longer be treated as setup work that happens before the real security tasks begin.
The official guide 2.0 (opens in a new tab) also adds a third subsection to data management: configuring and ingesting threat intelligence. It covers importing GTI or third-party indicators into the Entity Context Graph and using capabilities such as LiveHunt, Digital Threat Monitoring, Google Insights, and URL Scanning.
GTI becomes a cross-domain platform
The previous exam guide (opens in a new tab) already referred to GTI in several objectives. Version 2.0 goes further: its introduction now states that the exam assesses work in SecOps, SCC, and GTI. Candidates need more than awareness of threat-intelligence sources; they must select, integrate, evaluate, and apply indicators and tactics, techniques, and procedures.
In platform operations, candidates should evaluate surfaced IOCs and TTPs and integrate the ones that fit the use case. In threat hunting, they need to form hypotheses from attack paths, exposure context, and entity risk. In detection engineering, they must use Applied Threat Intelligence, the Emerging Threats Center, and curated detections. In incident response, they may need to respond to Digital Threat Monitoring alerts.
A useful study exercise should carry one scenario through all six domains. A threat-intelligence source contributes an IOC; ingestion adds it to context; a historical search finds related activity; a rule detects a behavior; a playbook enriches the case; and the team measures response speed and quality. That workflow is more valuable than memorizing separate product lists.
AI is now part of the measured work
Guide 2.0 names artificial intelligence in specific objectives. The audience profile now includes using AI for detection and response. Platform operations cover improving processes with AI and automation. Threat hunting includes prompting AI in natural language to conduct a hunt, while detection engineering includes using natural language to create detection rules.
Incident response explicitly refers to the Triage and Investigation Agent and the Gemini Summary widget in Google SecOps SOAR. Candidates must also analyze the agent’s results and understand how a playbook can invoke it. This does not turn the exam into a prompt-writing test. The practical skill is integrating generative assistance into a defensible process: preserving evidence, checking hypotheses, evaluating automation risk, and identifying decisions that require human review.
Detection objectives also include AI infrastructure risk. Candidates should be able to design rules that find posture or AI-infrastructure risks, then use asset and threat context to improve precision and dynamic risk scores. A good lab should include a misconfigured AI resource, privilege change, or rare behavior that produces multiple signals requiring correlation.
MCP and non-human identities
The access section is reframed around human and non-human identities. It covers authentication to security tools, including SCC, Google SecOps, and MCP, as well as API access for automation. That is an operational change: a connector, service account, or agent that triggers a response action needs limited, observable, and revocable access.
For this area, practice four decisions: identity type, IAM role, access conditions, and audit logs. Then add secret storage, key or token rotation, and separation between read, enrichment, and remediation privileges. MCP should not be studied as an isolated protocol. The guide places it inside a security architecture where tools and agents may call sensitive resources.
This objective complements the identity concerns covered in the SC-100 update on AI security. SC-100 remains focused on Microsoft security architecture, while the Professional Security Operations Engineer expects operational decisions within Google Cloud’s security ecosystem.
More precise data and detection objectives
The new guide makes ingestion objectives more concrete. It cites pipeline architecture, Bindplane for collecting Windows event logs, direct ingestion of SCC findings into SecOps, parser modifications, normalization, and cost management. It also distinguishes event logs from entity data used to enrich investigations.
In detection engineering, guide 2.0 asks candidates to analyze coverage against MITRE ATT&CK, build composite detections over longer periods, and configure SCC detectors for Cloud Run, containers, and virtual machines. It refers to data tables rather than only reference lists and introduces more explicit dynamic risk scoring.
A useful lab should begin with imperfect data. Take two sources with different schemas, define the important fields, validate a parser, enrich entities, and write a YARA-L rule. Then measure false positives, ATT&CK coverage, and ingestion cost. That end-to-end exercise is stronger preparation than a set of isolated product flash cards.
Exam format, price, and renewal
The official page still lists a two-hour exam with 50–60 multiple-choice questions, delivered remotely or at a test center. The listed fee is USD 200, plus applicable tax. There is no mandatory prerequisite, but Google recommends at least three years of security experience, including one year of hands-on work with Google Cloud security tools.
The standard certification is valid for two years. Google currently lists English and Japanese as available languages, but the update notice confirms November 24 only for the English version. Do not extend that date to Japanese without an official confirmation. Check both the language and guide version during scheduling.
Google also offers renewal through Google Skills. Eligible holders complete designated courses or skill badges during the final year of the active certification. The renewal path grants one year of validity. Exact requirements appear in the dashboard after the Google Skills and CM Connect accounts are linked.
A four-week preparation plan
The official Google Skills path (opens in a new tab) contains 13 activities across SecOps, SCC, and GTI. Use it as the spine of your plan, then add exercises that connect multiple domains.
Week 1: platform, access, and ingestion
Map SecOps, SCC, GTI, IAM, and telemetry sources. Create an access matrix for people, services, and automations. Then work through an ingestion pipeline, parser behavior, normalization, and cost controls.
Week 2: context and threat hunting
Build user, asset, and entity context; ingest safe test indicators; and state hunting hypotheses. Alternate structured queries with natural-language assistance, then verify results manually. Include one retrohunt and an analysis of entity risk scores.
Week 3: detection and coverage
Write YARA-L rules, test a composite detection, and map results to MITRE ATT&CK. Identify coverage gaps, false positives, and posture signals. Compare a manually designed rule with an AI-assisted proposal without assuming the latter is correct.
Week 4: response, automation, and observability
Design a playbook with enrichment, notification, escalation, and closure. Decide which steps can be automated and which need approval. Finish with health, latency, and quality metrics, then run a complete simulation from alert to retrospective.
For a broader view of Google Cloud’s agentic credentials, the Professional Agentic Architect timeline helps separate agent architecture from security operations. The two paths overlap around AI and governance, but they validate different jobs.
Should you test now or wait for guide 2.0?
If your English appointment is before November 24, prepare against the version shown when you schedule and confirm it in the testing portal. For an appointment on or after November 24, build directly from guide 2.0. Postponing is not automatically necessary: SecOps, SCC, GTI, YARA-L, IAM, and SOAR fundamentals remain relevant to both versions.
However, do not dismiss the update because the six domain names are unchanged. The higher data-management weight, explicit AI work, MCP, non-human identities, Entity Context Graph, and new investigation agents change the expected depth. The safest approach is to compare both official guides, close the gaps with hands-on exercises, and check the certification page again before booking.
