Hello, you are using an old browser that's unsafe and no longer supported. Please consider updating your browser to a newer version, or downloading a modern browser.
Everything you need to know about AI+ Security Strategist, the advanced exam in AI CERTs' AI+ Security track, as of October 2026. This guide covers the AT-2103 exam, the twelve domains, the name change, what AI CERTs does and does not claim about accreditation, and how the exam compares to SecAI+ and AAISM.
AI+ Security Strategist is the advanced, third exam in AI CERTs' AI+ Security track. It tests whether an experienced security professional can apply machine learning across network, endpoint, cloud, identity and IoT defense, and keep the AI doing that work secure.
The exam is AT-2103. You get 50 multiple-choice questions in 90 minutes, taken online through AI CERTs' own AI-proctored platform, and you need 70 percent (35 of 50) to pass. The twelve domains run from ML-based threat detection and adversarial AI to container security, blockchain, IAM and IoT. AI CERTs' scheme is plain about the scope. It says the exam validates advanced applied knowledge and does not independently certify leadership, governance or executive-level decision-making.
Until AI CERTs renamed its security track, this exam was called AI+ Security Level 3, and that is still the name on Training Camp's course page, on AI CERTs' certification body site and in the scheme documents. The code never changed. There are no mandatory prerequisites, though AI CERTs' course materials treat the two lower AI+ Security exams as the foundation. It is issued by AI CERTs, a New York company that also runs the AI CERTs store and training partner network.
Domains 1 to 3. Machine learning and deep learning for threat detection and response.
Domains 4 and 7. Adversarial attacks, red teaming AI, and secure AI architecture.
Domains 5, 6, 8 and 10. Network, endpoint, cloud and container, and identity.
Domains 9, 11 and 12. Blockchain, physical and IoT security, and scenario questions.
Four things worth knowing about AT-2103 before you weigh it against the other AI security credentials of 2026.
AI CERTs splits AI security into three exams. AI+ Security Practitioner (AT-2101) validates foundational knowledge, AI+ Security Expert (AT-2102) intermediate, and AI+ Security Strategist (AT-2103) advanced. All three use the same 50-question, 90-minute format with a 70 percent pass mark, so what changes is depth. AT-2103 assumes you know the security basics and spends its questions on engineering AI into real defenses.
The Strategist name suggests a leadership exam. The blueprint says otherwise. It covers intrusion detection, EDR, container security, MFA and smart contracts, and AI CERTs' own scheme says the exam does not independently certify leadership, governance or executive-level decision-making.
Twelve domains share 50 questions, so each one gets only about four on average. Expect breadth across the stack rather than detailed questions on any one tool. For how AI CERTs compares with other providers, see every major AI certification in 2026 and who it's for.
AI CERTs says it operates as a certification body under ISO/IEC 17024:2012, and its accreditation page says it is "actively pursuing ANAB Accreditation." That is a stated goal, not an accreditation, and AI CERTs does not claim one for AT-2103. It also claims no DoD 8140 listing, so check the current matrix at the DoD Cyber Exchange before you count on it for a cyber-coded position.
What it does offer is a short, inexpensive exam with a tight technical focus. The exam-only fee is $195. The tradeoff is a one-year validity, which means sitting an exam every year to keep the credential.
The certification, the AT-2103 exam, and how you keep it, as AI CERTs publishes them in October 2026.
AT-2103 is an engineering exam. Three other AI credentials cover what it leaves out: governing AI, managing an AI security program, and a vendor-neutral practitioner route that lasts three years instead of one.
For the policy and program side of AI. The IAPP Artificial Intelligence Governance Professional covers the EU AI Act, the NIST AI RMF and ISO 42001, with no prerequisites. Pick it if your work is privacy, compliance or legal rather than detection engineering.
For managers who run an AI security program. ISACA's Advanced in AI Security Management requires an active CISM or CISSP, which tells you the audience. AT-2103 is for the engineer who builds the controls an AAISM holder oversees.
CompTIA's CY0-001 exam, up to 60 questions in 60 minutes, valid for three years. It puts 40 percent of its weight on securing AI systems and adds a governance domain, where AT-2103 spends more of its time on ML inside network, endpoint and cloud defense.
AT-2103 closes out AI CERTs' three-exam security track. Most people reach it after the Practitioner and Expert material, then branch out depending on whether they want to govern AI, manage it, or add a longer-lived practitioner credential.
The two lower AI+ Security exams
The advanced AI security exam
Other AI credentials
Two questions to answer before you book: can you sit AT-2103, and should you pursue it rather than another AI credential. Here's a straight answer to both.
No mandatory prerequisites.
AI CERTs' scheme for this exam imposes no prerequisites related to education, experience or prior certifications, and any candidate may attempt it. You buy the exam, accept the candidate agreement, and test online through Proctor 365 with a government-issued ID for identity checks.
AT-2103 is written for experienced practitioners.
The scheme describes its target candidate as an advanced professional with substantial cybersecurity experience and foundational AI and ML knowledge, and it recommends programming skill in Python, R or similar. AI CERTs' program guide goes further and lists the two lower AI+ Security exams as the foundation for its course. You can skip AT-2101 and AT-2102. Skipping what they teach is a bad idea.
AI CERTs' scheme names the people it is written for, from security engineers and threat intelligence analysts to cloud security engineers and ML engineers who specialize in security. These six roles line up best with the twelve domains.
Builds and hardens the machine learning models and pipelines a security team depends on. Secure AI system engineering and adversarial AI are the two AT-2103 domains this job touches every week.
Uses machine learning to pick new threats out of large volumes of telemetry. The threat detection, deep learning and threat hunting objectives line up directly with this work.
Secures cloud workloads, containers and serverless functions. Domain 8 covers AI for cloud security, container security in orchestration platforms like Kubernetes, and DevSecOps.
Designs access controls around behavior analytics, adaptive MFA and zero-trust policy. Domain 10 reads like a job description for this role.
Protects industrial IoT, connected devices and smart infrastructure. Domain 11 covers smart cities, industrial IoT, autonomous vehicles and consumer devices.
Classifies malware samples and tunes endpoint detections. The endpoint domain covers machine learning for malware classification, polymorphic threats and EDR.
AT-2103 is easiest to judge next to the AI CERTs exam below it and the two AI security credentials people most often weigh it against. Shortest version: Strategist applies AI across the stack, SecAI+ secures and governs AI, AAISM manages the program.
| AI+ Security Strategist | AI+ Security Expert | CompTIA SecAI+ | ISACA AAISM | |
|---|---|---|---|---|
| Issuer | AI CERTs | AI CERTs | CompTIA | ISACA |
| Exam Code | AT-2103 | AT-2102 | CY0-001 | AAISM |
| Level | Advanced | Intermediate | Specialty (mid-level) | Advanced (management) |
| Exam | 50 questions, 90 min | 50 questions, 90 min | Up to 60 questions, 60 min | 90 questions, 150 min |
| Passing Score | 70% | 70% | 600 of 900 | 450 of 800 |
| Prerequisites | None mandatory | None mandatory | None (Security+, CySA+ rec.) | Active CISM or CISSP |
| Exam Cost | $195 exam only | $195 exam only | $298 voucher | $459 member, $599 non-member |
| Validity | 1 year, re-exam | 1 year, re-exam | 3 years (CE) | 30 CPEs over 3 years |
| DoD 8140 | No listing claimed | No listing claimed | Not listed | Not listed |
| Best For | Engineering AI into network, endpoint, cloud and IAM defense | Building intermediate AI security skills | Securing AI systems hands-on | Running an AI security program |
The AI CERTs exams are the cheapest and shortest here, but they expire after one year, while the CompTIA and ISACA credentials run on three-year cycles. Pick by the work you do and how often you are willing to retest.
Certification Matchups: the published facts, strengths and weak spots of 33 certifications, side by side.
Our AT-2103 boot camp runs five days, live and instructor-led, in person or online. As an AI CERTs Authorized Training Partner we teach from official AI CERTs courseware with videos, e-books and quizzes, plus hands-on labs and a capstone project in a provided environment. Your exam voucher and an exam retake are included, along with 180 days of one-on-one coaching after class.
Where AI CERTs sits among the AI credentials of 2026, how the CompTIA and ISACA options compare, and the agentic AI risks security teams are studying now.
A cert-by-cert rundown of the AI credentials employers ask about in 2026, from IAPP and ISACA to the cloud vendors and AI CERTs' AI+ family, with who each one is actually built for. Updated September 2026.
The other technical credential in AI CERTs' security family: what AI+ Ethical Hacker covers, who it suits, and the roles it points toward.
Why credentials backed by established bodies, regulatory pressure and employer demand are pulling ahead, with a close look at IAPP's AIGP and ISACA's three AI certifications.
SecAI+, AAISM and AAIA side by side, sorted by one question: do you implement AI security controls, manage the program, or audit it?
A straight look at who should take CompTIA's AI security exam and who should hold off. SecAI+ is the alternative most hands-on candidates weigh against AT-2103.
The ten risks the OWASP GenAI Security Project published in December 2025 for AI agents that plan, remember and call tools, and why agents needed their own list.
How to turn an official exam outline into a study plan. Handy for AT-2103, where the scheme lists twelve domains and you have to decide where your hours go.
AI CERTs' certification scheme organizes the AT-2103 exam into twelve domains. Most of them apply AI to one part of the security stack, from intrusion detection to IoT, and the last pulls everything together in scenario questions. Click any domain for what it covers.
Core AI and machine learning concepts applied to security, the main AI use cases in cybersecurity, and how to engineer AI pipelines for security data. It also covers the problems that come with applying AI to security, such as imbalanced datasets and attackers who target the models themselves.
Feature extraction from cybersecurity datasets, supervised learning for threat classification such as malware families, unsupervised learning for anomaly detection, and building detection systems that work in real time. AI CERTs' course material names Apache Kafka and Spark for streaming pipelines.
Convolutional neural networks for threat detection, recurrent networks and LSTMs for sequence data like logs and traffic, autoencoders for anomaly detection, and adversarial deep learning. AI CERTs lists network traffic classification, phishing detection and intrusion analysis as example uses.
How adversarial attacks on AI models work, the defenses against them (adversarial training and ensemble methods among them), adversarial testing and red teaming of AI systems, and engineering models that keep working when an attacker feeds them manipulated input.
AI-powered intrusion detection systems, AI for detecting distributed denial of service (DDoS) attacks, network anomaly detection, and designing secure network architectures with AI, including zero-trust designs and AI inside next-generation firewalls.
AI for malware detection and classification, including polymorphic malware, AI in endpoint detection and response (EDR), AI-driven threat hunting, and securing mobile and IoT devices with lightweight models that fit on constrained hardware.
Designing secure AI architectures, where cryptography fits in AI security systems, keeping models explainable and transparent enough to trust in a security decision, and tuning AI security systems so they perform under real-time load.
AI for securing cloud environments, container security inside orchestration platforms such as Kubernetes, securing serverless architectures, and bringing AI into DevSecOps practice and security testing.
How AI and blockchain work together, AI-based fraud detection on blockchains, smart contract security, and AI-enhanced consensus algorithms. If you have never worked with blockchain, budget extra time here, because the rest of the exam will not have prepared you for it.
AI for user behavior analytics, AI-driven multi-factor authentication (MFA), zero-trust architecture and role-based access control (RBAC). The theme is using AI to automate access decisions and catch unauthorized access early.
AI for securing smart cities, industrial IoT, autonomous vehicles, and smart homes and consumer IoT devices. AI CERTs' course outline also covers federated learning as a way to run security models across devices without pooling the data.
The September 2026 scheme calls this the Case Study domain, and older copies and AI CERTs' course pages call it the Capstone Project. On the exam it shows up as scenario-based multiple-choice questions that pull the other eleven domains together, from defining the problem to deploying and monitoring the system. There is no project to submit.
Domain names follow V3.0 of AI CERTs' certification scheme for AT-2103. AI CERTs has published two different sets of domain weights for this exam (the scheme's own figures add up to 110 percent), so this guide lists the domains without weights. Treat all twelve as testable and check the current blueprint on AI CERTs' site before you book.
The questions candidates ask most often when researching AI CERTs' AT-2103 certification.
AI+ Security Strategist is the advanced certification in AI CERTs' three-exam AI+ Security track, under exam code AT-2103. It tests whether experienced security professionals can apply machine learning and deep learning to threat detection and to network, endpoint, cloud, identity and IoT security, and keep the AI systems involved secure.
Yes. The AI CERTs store lists AI+ Security Strategist with exam code AT-2103 and the note "formerly known as AI+ Security Level 3." The certification body site at ai-certs.org and the scheme documents still use the older title with the same code. The two lower exams were renamed too, to AI+ Security Practitioner (AT-2101) and AI+ Security Expert (AT-2102).
The exam has 50 multiple-choice, single-response questions and runs 90 minutes, which includes 5 minutes for the candidate agreement and 5 minutes for a system tutorial. You need 70 percent, or 35 correct answers, to pass. You take it online through Proctor 365, AI CERTs' own AI proctoring platform. The case study questions are scenario-based multiple choice inside those 50, and there is no separate project to submit.
There are twelve: Foundations of AI and Machine Learning for Security Engineering, Machine Learning for Threat Detection and Response, Deep Learning for Security Applications, Adversarial AI in Security, AI in Network Security, AI in Endpoint Security, Secure AI System Engineering, AI for Cloud and Container Security, AI and Blockchain for Security, AI in Identity and Access Management, AI for Physical and IoT Security, and a Case Study on engineering AI security systems. AI CERTs has published two different sets of weights for them, so confirm the current blueprint before you plan your study time.
No. The AT-2103 certification scheme says there are no mandatory prerequisites related to education, experience or prior qualifications, and any candidate may attempt the exam. AI CERTs' program guide does list the two lower AI+ Security exams as the foundation for its course, and the exam is written for people with substantial cybersecurity experience, so most candidates should know that material before they book.
As of October 2026, the AI CERTs Candidate Handbook lists the exam-only fee for AT-2103 at $195 USD and the recertification exam at $95. The AI CERTs store sells the self-paced AI+ Security Strategist course with the exam for $495. Training Camp's AT-2103 boot camp includes the exam voucher in the program fee.
Yes. The AI CERTs Candidate Handbook says every exam voucher includes one free retake, which you can take after a 7-day cooling period. Vouchers, retake included, expire 6 months after purchase. After the free retake you need to buy another voucher, and AI CERTs recommends about 30 days of preparation before a third attempt.
One year from the date it is issued. The scheme says the only way to recertify is to pass a proctored multiple-choice re-examination before the certificate expires, and the Candidate Handbook lists the recertification exam at $95. If it lapses and you never requested a grace period, you have to go through the full certification process again.
AI CERTs says it operates as a certification body under ISO/IEC 17024:2012, and its accreditation page says it is "actively pursuing ANAB Accreditation." It does not claim ANAB accreditation for AT-2103. If an employer or contract asks for an accredited credential, don't assume this one qualifies until AI CERTs announces otherwise.
AI CERTs does not claim a DoD 8140 listing for AT-2103. If you need a credential for a cyber-coded position, check the current qualification matrix on the DoD Cyber Exchange before you rely on this one. See the current DoD 8140 work role paths.
Take AI+ Security Strategist if you want an advanced exam on applying AI across network, endpoint, cloud, IAM and IoT defense, and you are fine renewing every year. Take SecAI+ if you want CompTIA's vendor-neutral AI security exam, which puts 40 percent of its weight on securing AI systems and stays valid for three years. They overlap less than the names suggest.
It lines up with AI security engineer, threat intelligence analyst, cloud security engineer, IAM engineer, IoT security specialist and malware analyst roles. AI CERTs' scheme also names penetration testers, incident response teams and machine learning engineers who specialize in security among its target candidates. Treat it as proof of a skill set that sits on top of experience, not as an entry ticket on its own.
It is worth it if you already work in security engineering and want a quick, low-cost way to show you can apply machine learning across the stack. It is a weaker pick if you need DoD 8140 approval, a credential that lasts longer than a year, or third-party accreditation. In those cases SecAI+, AAISM or AIGP may serve you better.
Whether you're choosing between AT-2103 and SecAI+, wondering if you should take the Practitioner and Expert exams first, or planning AI security training for a team, tell us where you are and we'll help you map out the right path.