Is the AIGP Exam Hard? Format, Passing Score & Study Time

Last reviewed: 2026-07-05 · By Victor Humenhuk (AIGP certified, 2026)

The short answer

The AIGP is moderately hard - harder than its reputation, easier than people fear. I passed it in 2026, after CIPP/E and CIPP/US, and my honest read is this: no single part of it is difficult, but the breadth is. In one sitting you're tested on machine-learning fundamentals, harms and ethics taxonomies, governance frameworks like NIST AI RMF and ISO 42001, the EU AI Act in real detail, how GDPR and IP law collide with AI, and lifecycle governance from planning through to decommissioning. Nobody arrives knowing all of that.

It's a genuinely hybrid exam. Privacy lawyers cruise through the legal modules and then stumble on transformer architectures and overfitting. Engineers breeze past the technology and then get tangled in EU AI Act roles. Whichever side you come from, roughly a third of the syllabus will be new - plan for that and it's very passable.

What makes it hard - and what makes it manageable

What makes it hard:

• Breadth over depth. Eight modules spanning technology, ethics, frameworks, regulation, existing law, development and deployment. Cramming one area well won't carry you.
• Precision on the EU AI Act. Questions distinguish providers from deployers, prohibited from high-risk, and one high-risk requirement from another. Vague familiarity fails here; exact recall passes.
• Look-alike vocabulary. Transparency vs explainability, fairness vs bias types, drift vs degradation, the different learning paradigms - the wrong options are always the neighbouring concept.
• Case-study questions. The exam includes scenario items where several answers are defensible and you must pick the best one, which punishes surface-level reading.

What makes it manageable:

• It rewards structured study. The body of knowledge is well defined; there are no trick topics outside it. If you cover the eight modules, you've covered the exam.
• No penalty for guessing, no section minimums. Your score is based only on correctly answered scored questions, so a weak module can be carried by strong ones.
• Concepts repeat. The AI lifecycle, risk thinking and the harms vocabulary recur across modules - the second half of your revision reinforces the first half for free.
• No prerequisites. IAPP designed it to be passable without prior certification or experience, and that shows in the question style: it tests understanding, not war stories.

Format and passing score

The verifiable facts, per iapp.org:

Two things people search for that don't exist: IAPP does not publish a pass rate for the AIGP (or any of its exams - the FAQs, candidate handbook and certification pages contain no pass-rate figures at all, and IAPP says results are kept for six months and then destroyed). And IAPP won't tell you how many questions you got right - you only ever see the scaled score. Earlier candidate handbooks also stated that some questions are unscored items that don't count toward your result; the current split isn't published on live pages, so check the current candidate handbook at iapp.org. The practical takeaway: an absurdly hard question may not even count, so answer it and keep moving.

How long should you study?

This is my judgement from doing it, not an official figure.

Whatever your track, the last week should be mostly practice questions, not re-reading. My study guide lays this out as a concrete week-by-week plan.

The traps that catch people

These are the failure modes I'd warn a friend about:

1. Treating it as a law exam or a tech exam. It's both. The most common mistake is over-studying your home turf and skimming the unfamiliar third of the syllabus - which is exactly where the exam will find you.
2. Fuzzy EU AI Act roles and tiers. Provider vs deployer vs importer vs distributor, and prohibited vs high-risk vs limited-risk classification, are tested with scenarios where the wrong answers are the adjacent role or tier. Learn them to the point of instant recall.
3. Framework blur. NIST AI RMF, ISO 42001, the OECD principles and the trustworthy-AI attributes overlap just enough to be dangerous. Know which framework owns which concept and vocabulary.
4. Vocabulary near-misses. A large share of questions are really definition questions wearing a scenario costume. If you can't cleanly separate explainability from transparency, or the learning paradigms from each other, you'll donate easy marks.
5. Panicking over strange questions. Some questions are simply hard - and per earlier candidate handbooks, some may be unscored items that don't count toward your result at all. Fixating on one costs you time you need for the case studies. Flag, guess (there's no penalty), move on.
6. Misreading 300 as 60%. IAPP says outright that the passing score doesn't map to a percentage. Don't build your practice targets around a myth - aim comfortably high on practice runs instead.

How to prepare

My whole preparation system is on this site, and the core of it is free:

• Study the notes. The AIGP study guide maps all eight modules of the body of knowledge to free plain-English notes, with a week-by-week plan built around active recall.
• Test yourself early and free. The 28 free practice questions will tell you which modules need work before you've sunk weeks into the wrong ones.
• Drill until it's boring. The full 229-question bank (one-time unlock, lifetime access) includes exam-style case studies. When you're consistently above 80% and can explain why each wrong option is wrong, book the exam.
• Get the exam-day tactics right. How to pass the AIGP covers pacing across 100 questions, using the break, and handling the case-study blocks.

So: is the AIGP hard? It asks for genuine breadth and precise recall, and it will punish a one-sided background left unaddressed. But it's a fair, well-defined exam with no prerequisites and no mystery topics - study all eight modules properly, practise until your misses run dry, and you should walk out with a pass on the screen.