Two people read this page. A hirer looking at a line on a CV and wondering what it means, and a specialist deciding whether a course is worth the money. The honest answer is the same for both, so here it is once.
Our position
A certification is never enough to be admitted here. Not one of them, not five. The rubric requires evidence of work that was checked, and a certificate is evidence that an exam was passed.
A certification does count as supporting evidence beside work we can check. Where somebody has shipped a governed AI system and holds a governance credential, the credential tells us the vocabulary is shared and the framework is not being invented on the spot.
The certifications that carry the most weight are the ones that gate on experience before you may sit them. ISACA requires an existing audit credential for AAIA. That entry bar does more work than the exam does.
We will not tell a specialist to go and get a certificate to join. It would cost them money, it would not change our decision, and telling a work-seeker to spend money to become placeable is exactly the behaviour UK recruitment rules exist to prevent.
Worth something
A buyer reasonably treats this as evidence. It is either hard to pass without the skill, or it gates on experience before you can sit it.
Working familiarity with building, deploying and operating machine learning workloads on AWS, including the pipeline and monitoring services rather than only the model.
Does not prove
That the holder has run anything at scale, or that they can tell a good architecture from a defensible one. It is scoped to one cloud, and the exam rewards recognising the AWS answer rather than the right answer.
AWS states the current MLA-C01 exam can be taken in English until 28 September 2026, with registration for the updated MLA-C02 opening 1 September 2026. Read on the AWS certification page, 8 August 2026.
Practical assembly of AI features from Azure services, including the language and search components most retrieval systems are built on in Microsoft shops.
Does not prove
That the holder can evaluate whether a retrieval system is returning the right thing. Assembly and evaluation are different skills, and the second one is the one that is scarce.
International Association of Privacy Professionals
Proves
Command of the vocabulary, the emerging legal landscape and the accepted structures of AI governance, tested by a three-hour, 100-item exam including case studies.
Does not prove
That the holder has ever made a governance decision that cost somebody something. Governance is easy to describe and hard to enforce, and the exam can only test the first.
IAPP states the certification term is two years and requires 20 continuing education credits to maintain. Read on the IAPP certification page, 8 August 2026.
Auditing capability applied to AI systems, from someone who already holds an audit credential. The eligibility bar is the interesting part: ISACA requires CISA or an equivalent accounting or internal audit designation before you can sit it.
Does not prove
Technical depth. An auditor who can interrogate a model risk register is not the same as somebody who can tell you whether the evaluation set was any good.
ISACA requires 10 hours of AI-specific continuing education annually, and gives five years from passing to apply. Read on the ISACA credentialing page, 8 August 2026.
International Association of Privacy Professionals
Proves
Working knowledge of European data protection law, which is the ground most AI governance work actually stands on.
Does not prove
Anything AI-specific. It predates all of this, and that is precisely why it is worth more than it looks: a governance lead who does not understand the data law underneath is going to be wrong in expensive ways.
Genuine hands-on offensive capability, assessed by a long practical examination rather than by multiple choice. One of very few certifications anywhere that is hard to pass without the skill.
Does not prove
Model red teaming. It is a systems and network qualification. Somebody holding it can probably learn to attack a model; it is not evidence that they have.
Familiarity with the AI management system standard and how an audit against it runs. Increasingly the thing a procurement team asks for by name.
Does not prove
Consistency. The training is delivered by many different bodies to varying depth, so the name of the training provider tells you more than the certificate does.
The standard itself is published by ISO. The training and certification around it are not, so check who awarded it before treating two of these as equivalent.
A shared vocabulary. Useful for a non-technical leader who has to sit in the room and follow the argument.
Does not prove
Anything at all about building or assessing AI systems. BCS states there are no entry requirements, and the certificate is honest about being a foundation.
BCS runs it as a pathway: four awards, one in each of four areas, make up the Foundation Certificate. Read on the BCS certification page, 8 August 2026.
Work that happened, that somebody else can confirm happened, and that the person can still explain when you push on it. The rubric sets out how that is assessed and what gets an application declined.
13 certifications listed. Every entry read on the issuer’s own site on 8 August 2026. Where a fact is time-bound, the date it was read is stated beside it.