12 AI Certifications in 2026: Which One Fits Your Job, and What to Take First
AI certifications went from a handful of fundamentals exams to a crowded field in about eighteen months, and 2026 reshuffled it again: exams were retired, replaced, rebuilt for agents or released in beta. If you're trying to pick one, the brand on the badge matters less than one question: what do you actually do with AI at work? Here are the twelve credentials worth knowing, who each one is built for, which ones need another cert first, and the order we'd take them in. And as of this month, we teach all twelve.
Start from your job, not the vendor
Every AI certification lands in one of five lanes. Pick the lane that matches the work on your desk, then pick the vendor that matches the platform your company already runs on.
You use AI but don't build it
Analysts, product owners, managers, sales engineers. Start with AWS Certified AI Practitioner or Microsoft AI-901.
You build AI apps or ML pipelines
Take the builder exam for the cloud you work in: AI-103 (Azure), MLA (AWS), Professional ML Engineer (Google) or NCA-GENL (NVIDIA).
You defend systems
Security analysts and engineers start with CompTIA SecAI+. Security managers who already hold CISM or CISSP go to ISACA AAISM.
You audit or govern
IT auditors with CISA take ISACA AAIA. Privacy, legal, compliance and risk teams take IAPP AIGP.
You run AI projects or programs
Project managers take PMI-CPMAI. Leaders running AI across a whole organization look at EC-Council C|AIPM.
All twelve, side by side
| Certification | Body | Lane | Exam | Prerequisite | Renewal |
|---|---|---|---|---|---|
| AI Practitioner (AIF-C01) | AWS | Foundations | 65 questions, 90 min, pass 700/1000 | None | 3 years |
| Azure AI Fundamentals (AI-901) | Microsoft | Foundations | Pass 700; length not published | None | Doesn't expire |
| ML Engineer – Associate (MLA) | AWS | Build | MLA-C02 beta: 85 questions, 170 min; pass 720 | None (1 yr SageMaker recommended) | 3 years |
| Professional ML Engineer | Google Cloud | Build | 50–60 questions, 2 hours | None (3+ yrs recommended) | 2 years |
| AI Apps and Agents Developer (AI-103) | Microsoft | Build | 120 min, pass 700 | None | Every year, free online |
| Generative AI LLMs (NCA-GENL) | NVIDIA | Build | 50–60 questions, 1 hour | None | 2 years |
| SecAI+ (CY0-001) | CompTIA | Secure | Up to 60 questions incl. performance-based, 60 min, pass 600/900 | None (Security+ level recommended) | 3 years (CE) |
| AAISM | ISACA | Secure | 90 questions | Active CISM or CISSP | 10 CPE/yr, 30 per 3 yrs |
| AAIA | ISACA | Audit | 90 questions, 150 min, pass 450/800 | Active CISA (or a listed audit/accounting designation) | 10 AI CPE/yr, 30 per 3 yrs |
| AIGP | IAPP | Govern | 100 questions incl. case studies, 2 h 45 min | None | 20 CPE per 2 yrs |
| PMI-CPMAI | PMI | Manage | 120 questions (20 unscored), 160 min | PMI's 21-hour prep course | 30 PDUs per 3 yrs |
| C|AIPM (312-41) | EC-Council | Manage | 100 questions, 3 hours | Official training, or 2 yrs experience | Per EC-Council policy |
The 2025–2026 shake-up, in order
If a study guide, video course or practice test was made before 2026, check the exam code on it first. Several of these exams were replaced outright.
- Sep 30, 2025PMI-CPMAI launches
Replaces the older CPMAI v7 and adds a required 21-hour prep course. - Feb 2026New AIGP body of knowledge
IAPP's new study guide takes effect, organized into four domains. - Feb 17, 2026CompTIA SecAI+ launches
The first vendor-neutral exam built specifically for securing AI. - Jun 30, 2026Microsoft AI-900 retires
AI-901 is now the Azure AI Fundamentals exam. The older AI-102 developer exam has also been retired, and AI-103 is the new AI apps and agents developer exam. - Sep 28, 2026Last day for AWS MLA-C01 in English
The MLA-C02 beta began delivery on September 29, with no general-availability date announced yet. - 2027More coming
ISC2 is piloting a full AI security certification in late 2026, planned to be operational in 2027. ISACA has also announced a third AI credential, AAIR (Advanced in AI Risk).
Start with AI
Build your AI foundation with a credential that fits your cloud ecosystem. For most people this is the first AI certification, and it's the start of the foundation track.
AWS Certified AI Practitioner
AIF-C01 · Foundational
- For people who use AI and ML on AWS without necessarily building it.
- 65 questions (50 scored), 90 minutes, pass at 700 on a 100–1,000 scale.
- No prerequisite; AWS suggests up to six months of exposure to AI/ML on AWS.
- Valid three years. Passing the AWS ML Engineer Associate also renews it.
- Exam fee: $100, AWS's standard U.S. price.
Azure AI Fundamentals
AI-901 · Fundamentals
- Replaced AI-900, and it's more hands-on: more than half the exam is building in Microsoft Foundry.
- Skills: AI concepts and capabilities 40–45%; implementing AI solutions in Microsoft Foundry 55–60%.
- Pass at 700. Python and Azure familiarity are recommended.
- Fundamentals certifications don't expire.
Build AI
For developers, ML engineers and teams turning AI models into real applications. Choose the technology stack you use: here the cloud matters. Hiring managers want the badge for the platform they run, so don't take the Google exam if your company is all-in on Azure.
AI-103
Azure AI Apps and Agents Developer Associate
- Python developers building agents and AI apps in Microsoft Foundry.
- Biggest domain: generative AI and agentic solutions, 30–35%.
- 120 minutes, pass at 700. Expires every year; renewal is a free online assessment.
ML Engineer – Associate
MLA-C02 (beta) · MLA-C01 (ending)
- Engineers putting ML workloads into production on AWS.
- MLA-C02 adds foundation models, Amazon Bedrock, RAG, agents and responsible AI.
- Beta: 85 questions in 170 minutes. Pass at 720. Valid three years.
Professional ML Engineer
Professional
- Build, scale, serve and monitor traditional and generative models on Google Cloud.
- Updated for Google's move from Vertex AI to Gemini Enterprise Agent Platform.
- 50–60 questions, two hours. No coding on the exam, but expect to read Python and SQL.
NCA-GENL
Generative AI LLMs · Associate
- An entry point for people developing and maintaining LLM applications.
- Core ML and AI knowledge 30%, software development 24%, experimentation 22%, data analysis 14%, trustworthy AI 10%.
- 50–60 questions in one hour. Valid two years.
Watch the renewal clock on builder certs
AI-103 expires every year and AWS rebuilt its ML engineer exam about two years after launch. That isn't a reason to skip them, it's a reason to take them when you need them for a role, not two years before.
Secure AI
Defend intelligent systems and manage the security risks AI introduces. It's the fastest-growing lane: Every company that ships a chatbot just created a new attack surface: prompt injection, poisoned training data, leaked model outputs and agents with too much access. Security teams are being asked to own it, and these two credentials prove it: one for technical defense, one for security leadership.
CompTIA SecAI+
CY0-001 · launched February 17, 2026
- For working security practitioners who defend AI systems and use AI in security operations.
- Up to 60 questions, including performance-based items, in 60 minutes. Pass at 600 on a 100–900 scale.
- CompTIA recommends 3–4 years in IT and 2+ in hands-on security, at the level of Security+, CySA+ or PenTest+.
AAISM
Advanced in AI Security Management
- For security managers who already run a security program.
- You must hold an active CISM or CISSP first.
- 90 questions. AI technologies and controls 38%, AI governance and program management 31%, AI risk management 31%.
Govern AI
Build trust through responsible AI, strong governance and independent audit. Accountability matters more as AI scales, and these are the people who say "prove it."
AAIA
Advanced in AI Audit
- For IT auditors and advisors who audit AI systems.
- Requires an active CISA, or a listed designation such as CIA or CPA held by someone in IT audit or advisory.
- 90 questions, 150 minutes, pass at 450 on a 200–800 scale.
- AI operations 46%, AI governance and risk 33%, AI auditing tools and techniques 21%.
AIGP
Artificial Intelligence Governance Professional
- For privacy, legal, compliance and risk teams.
- Four domains: foundations of AI governance; laws, standards and frameworks; governing AI development; governing AI deployment and use.
- 100 questions with case studies, 2 hours 45 minutes plus a 15-minute break.
- 20 CPE credits every two years.
Lead AI
Turn promising AI initiatives into measurable business outcomes, from individual projects to enterprise impact.
PMI-CPMAI
Certified Professional in Managing AI
- For anyone running AI projects, using PMI's six-phase CPMAI method. Tool-agnostic and no experience required.
- You must finish PMI's 21-hour exam prep course before you can schedule.
- 120 questions (20 unscored) in 160 minutes. 30 PDUs every three years.
- Pairs naturally with the PMP, and the PDUs count toward both.
C|AIPM
Certified AI Program Manager · 312-41
- For leaders running AI strategy, adoption, governance and ROI across an organization. No coding background required.
- Readiness, use-case selection, strategy, change management, governance and measuring value.
- 100 questions in three hours. The cut score is set per exam form, between 60% and 85%.
Notice where the weight sits: more than half of CPMAI is business need and data, not models. That matches what goes wrong on real AI projects. They rarely fail because the model was bad. They fail because nobody checked whether the data existed.
Our takeFive things the brochures won't tell you
- Two of these are locked. AAISM needs an active CISM or CISSP, and AAIA needs an active CISA. If either one is your goal, the real first step is the base cert, and it's the longer of the two.
- Pair one technical and one governance credential. SecAI+ with AIGP, AI-103 with CPMAI, AWS MLA with AAIA. Teams are hiring people who can both build it and defend it in a review.
- Check the exam code on every study resource. AI-900, AI-102, MLA-C01 and CPMAI v7 are all gone or going. Old practice tests teach the wrong exam.
- Don't buy a builder cert to "get into AI". If you don't write Python, the associate developer exams will hurt. Start with a foundations exam and a few weeks of real practice.
- Security people already have a head start. If you hold Security+, CySA+ or CISSP, SecAI+ builds on what you know. You don't start over, you add the AI attack surface to it.
Your AI path
The right certification starts with the work you want to do.
Start with AI
- AWS AI Practitioner
- Microsoft AI-901
Build AI
- AWS ML Engineer
- Google Professional ML Engineer
- Microsoft AI-103
- NVIDIA NCA-GENL
Secure and govern AI
- CompTIA SecAI+
- ISACA AAISM
- ISACA AAIA
- IAPP AIGP
Lead AI
- PMI-CPMAI
- EC-Council C|AIPM
What to take first, by role
| If you are… | Take first | Then |
|---|---|---|
| New to AI, any role | AWS AI Practitioner or AI-901 | The builder exam for your cloud |
| Security analyst or engineer | Security+ (if you don't have it), then SecAI+ | CISSP or CISM, then AAISM |
| Security manager with CISM or CISSP | AAISM | AIGP for the regulatory side |
| IT auditor | CISA | AAIA |
| Privacy, legal or compliance | AIGP | AAIA if you audit, CRISC if you own risk |
| Developer on Azure, AWS or Google | AI-103, MLA or Professional ML Engineer | NCA-GENL for the model-side depth |
| Project or program manager | PMI-CPMAI | PMP, or C|AIPM if you run a whole program |
Start your AI path with us
We now teach all twelve of these certifications as instructor-led training camps, in Sterling, Atlanta and Burbank or live online, with classes scheduled through 2027. The base certifications the advanced AI credentials require (Security+, CISSP, CISM and CISA) run on our live schedule too. Not sure which one fits? Tell us what you do, and we'll point you to the right one.
Call 855-868-4467 or email website@thinqtanklearning.com
— Scott
- AWS Certified AI Practitioner
- AWS Certified Machine Learning Engineer – Associate
- Microsoft AI-901 study guide
- Microsoft Azure AI Apps and Agents Developer Associate (AI-103)
- Google Cloud Professional Machine Learning Engineer
- NVIDIA Generative AI LLMs Associate
- CompTIA SecAI+
- ISACA AAISM
- ISACA AAIA
- IAPP AIGP
- PMI-CPMAI
- EC-Council C|AIPM
- ISC2 AI security certification
Exam details are from each certifying body's published pages as of October 12, 2026, and change often. Confirm the current version before you schedule.