CompTIA · CY0-001
CompTIA SecAI+ is the first CompTIA Expansion Series certification. It validates the skills of mid-career IT and security professionals who protect AI systems and use AI in security operations.
Five minutes, free, no account.
Five minutes, free, no account — find out whether you’d pass.
Take the free diagnostic4 domains, each weighted by its share of the exam. Weights from the CY0-001 Exam Objectives v4.0 (V1; effective Feb 17, 2026).
| Domain | Weight |
|---|---|
| Securing AI Systems | 40% |
| AI-assisted Security | 24% |
| AI Governance, Risk, and Compliance | 19% |
| Basic AI Concepts Related to Cybersecurity | 17% |
At 40% of the exam, domain 2 is the largest section by a clear margin; the next largest is AI-assisted Security at 24%. It covers threat modelling, security controls, access controls, data security controls, monitoring and auditing, and reading the evidence of an attack. Four of its six objectives open with "Given a scenario", so most of this domain is applied judgement rather than definitions.
Many questions offer four controls that are all real and all reasonable, and the deciding fact is where the control is enforced: at the model, at the gateway or API, at the data, or at the identity. A rate limit, an output guardrail, a scoped token and a retrieval-time entitlement can all look like the answer to "stop this" until you ask which layer the stated requirement actually names.
Data poisoning, backdoors and evasion are not mutually exclusive: a backdoor is often implanted through poisoned training data. Jailbreaks are a kind of direct prompt injection, not something separate from it. Model extraction, model inversion and membership inference all steal something different. Wrong answers usually come from treating overlapping categories as rivals.
Data at rest, in transit and in use carry different controls, and the exam expects you to pick from the right set. "Encrypted while the model is computing on it" points at confidential computing or homomorphic encryption, not at TLS or disk encryption. Anonymisation, classification, redaction, masking and minimisation are also tested against each other, and they are not interchangeable.
CompTIA has published no SecAI+ simulator, and the performance-based items are ordering, matching and drop-down tasks rather than a virtual machine. Expect to sequence a life-cycle or incident-response flow, match controls to risks or resources to purposes, and fill in a short decision. Budget for them: 60 minutes for up to 60 questions leaves little slack.
The EU AI Act is binding law in its territory. The OECD AI Principles are voluntary recommendations with no supranational enforcement. ISO's AI management-system standard is certifiable; other standards in the same family are guidance and are not. The NIST AI Risk Management Framework is voluntary and has no certification at all. Questions turn on exactly these distinctions, and on whether a duty falls on a provider or a deployer.
OWASP's LLM Top 10 covers risks in LLM applications; its Machine Learning Security Top Ten is a separate, still-draft project covering the ML pipeline. MITRE ATLAS is a knowledge base of adversary tactics and techniques against AI. The MIT AI Risk Repository is a taxonomy of documented risks. The CVE AI Working Group defines what counts as a CVE-able AI vulnerability and how disclosure works; it does not assign CVE IDs. Scope, not topic, is what separates them.
Domain 3 repeatedly asks you to separate where an analyst works from what the AI does: an IDE plug-in, a browser plug-in, a CLI tool, a chatbot and an MCP server are integration surfaces, while vulnerability analysis, anomaly detection, pattern recognition and summarisation are use cases. The automation questions then ask how much autonomy is appropriate, and whether a documented human approval gate is required before the action runs.
SecAI+ (CY0-001) is CompTIA's certification for securing AI systems and using AI in security work. It is the first certification in CompTIA's Expansion Series, and it is aimed at mid-career IT and security professionals rather than beginners.
One exam: CY0-001.
Security practitioners who already work in the field and now have AI systems to protect, or who are being asked to use AI tools in security operations. CompTIA recommends 3 to 4 years of IT experience including about 2 years of hands-on cybersecurity experience.
No. There are no enforced prerequisites, so anyone can book the exam. CompTIA suggests Security+, CySA+ or PenTest+ or equivalent knowledge as helpful preparation.
A maximum of 60. CompTIA does not publish how many questions come from each domain or how many are unscored.
60 minutes, in one sitting with no scheduled breaks. That is a minute a question if you get the full 60, so pace matters more here than on longer CompTIA exams.
600 on a scale of 100 to 900.
No. The score is scaled, and CompTIA does not publish how raw answers convert to it, so 600 cannot be turned into a percentage of questions you must get right.
CompTIA describes the exam as a mix of multiple-choice and performance-based questions. The multiple-choice items are single response.
Not confirmed. The formats CompTIA publishes for SecAI+ are single-response multiple choice and performance-based, so multiple-response items should not be assumed.
CompTIA has published no SecAI+ simulator, so expect ordering, matching and drop-down tasks rather than a virtual machine: sequencing a life-cycle or response flow, matching controls to risks, or completing a short decision.
Securing AI Systems, at 40%. It is followed by AI-assisted Security (24%), AI Governance, Risk, and Compliance (19%) and Basic AI Concepts Related to Cybersecurity (17%).
No. The exam is about securing AI systems and using AI tools, not about developing them. You are expected to understand AI types and training techniques well enough to reason about their security consequences, not to implement them.
There is no subnetting-style calculation requirement. Where numbers appear they are usually thresholds and units in a scenario, such as a token budget, a rate limit or a size quota, and the question is which control the stated unit points to.
Yes. Prompt injection appears across the securing-AI and AI-assisted-security domains. Note that jailbreaks are a kind of direct prompt injection rather than a separate category, and that indirect prompt injection arrives through content the model retrieves rather than from the user.
The threat-modelling resources named in the objectives, including the OWASP Top 10 for LLM Applications, the OWASP Machine Learning Security Top Ten, MITRE ATLAS, the MIT AI Risk Repository and the CVE AI Working Group, plus the governance instruments in domain 4: the EU AI Act, OECD AI Principles, ISO's AI standards and the NIST AI Risk Management Framework.
Yes. The exam names tool classes and integration surfaces, such as IDE plug-ins, browser plug-ins, CLI tools, chatbots and Model Context Protocol servers, rather than particular products, so there is no product UI to memorise.
No. It is an Expansion Series certification that sits alongside the core pathway and adds AI-specific coverage, rather than replacing a level in it.
Yes. You can sit it at a Pearson VUE test center or online through OnVUE. Online you get one display and a built-in digital whiteboard, and physical notes are not allowed.
English and Japanese.
The price varies by country and changes over time, so check the current price with CompTIA before you book. Every attempt is charged at the full price.
You can book a second attempt straight away. Before your third and every later attempt you must wait at least 14 days. Each attempt costs the full exam price.
No. CompTIA does not publish pass rates, and figures quoted on blogs are estimates. Because the exam is new, there is also little reliable candidate data behind them.
Three years. It is renewed through CompTIA's continuing education programme with 15 CEUs and a renewal fee.
CY0-001 launched on 17 February 2026. CompTIA says exams usually retire about three years after launch, so retirement is estimated for 2029. No date has been announced.
The beta exam. CY0-001 is the live V1 exam; CY1-001 was the beta that preceded it, and study material referring to the beta code describes the same certification.
Last updated . CompTIA SecAI+ is a trademark of CompTIA. Enmaster is not affiliated with or endorsed by CompTIA.
5 minutes · no account, no card.
© 2026 Enmaster.
