AI certifications: where to start

By Miguel Maio · Published 18 October 2026

Since February 2025, having AI literacy stopped being an extra: it became a legal obligation for whoever hires you. Almost no junior knows this, and it's the biggest advantage you can bring to an interview.

Let me explain. Article 4 of the EU AI Act requires organisations to ensure that the people dealing with AI systems have sufficient literacy to do so. For a recruiter, that translates into a problem to solve. A junior who already brings that literacy solves it for free. You stop being the one who needs training and become the one who brings it.

I've taught since 2009 and I see the mistake repeat itself: people stacking AI certificates like collecting tattoos, waiting for the badge to do the work. It doesn't. The certificate opens the ATS door, because "GenAI" is now a screening keyword. What gets you through the door is the portfolio.

So this article isn't a catalogue. It's a path, with one measuring stick: what each certification proves to whoever decides.

The measuring stick: what the badge proves, not the badge

Not all AI certifications are the same thing, and treating them as equal is the first mistake. There are three types, and the difference is in who guarantees what.

The completion certificate says you finished a course. It proves you sat down and did the work. It counts as an ATS keyword and a sign of initiative, little more.

The proctored exam (an AI-900, say) says you passed an independent assessment. It proves knowledge verified by a third party, and so it carries more weight in a corporate context.

And there's the portable credential, tied to an issuer with reputation (ETS, Pearson, Coursera). It proves the same, but travels better between platforms and employers.

Keep this stick. You'll need it at every line of the path.

The path, in order of return

Start with what pays back fastest and climb as you want weight.

First, the free and fast base. OpenAI Academy has free courses with a completion certificate, good for the fundamentals and for the first "GenAI" on your CV. Watch out for a confusion going around: the formal OpenAI Certification (AI Foundations) is still in pilots with employers and institutions, with no public enrolment. Don't promise it on your CV as if you had it: you can't take it yet.

Then, Google AI Essentials. Short, built for productivity with AI at work, free to audit on Coursera. It's the piece that speaks the generalist market's language. If you want to go deeper, there's Google's AI Professional track.

Next, Career Essentials in Generative AI, from Microsoft with LinkedIn. Free, with a certificate, and with the practical advantage of landing straight on your LinkedIn profile, which is the database where recruiters look for you.

For the more technical, Anthropic Academy and the DeepLearning.AI short courses (prompt engineering, LLMs) are free and serious. And Hugging Face Learn, for those who want to touch open-source models and not just interfaces.

At the top, when you want exam weight, Microsoft AI-900 (Azure AI Fundamentals). It's the only one on this list with a paid, proctored exam, entry-level, and the one that counts most on a corporate CV. It's not where you start. It's where you arrive when you want a credential an IT department recognises.

The table, to decide at a glance

The formal OpenAI Certification is left out on purpose: it isn't public yet.

The unbreakable limit: the certificate is the door, not the house

Here's the part nobody tells you. You can have all five certificates and still not get past the first interview. Because what impresses an employer in 2026 isn't "I did an AI course". It's showing a system that works: a workflow of yours, with quality control and numbers.

You can have the five certificates, the profile full of badges and the meeting booked. It all goes beautifully, until they ask you to show one thing you've actually done.

A concrete example, worth more than the five badges together: an AI-assisted content pipeline, documented. Intent clustering with an LLM over your Search Console data, briefs with your own prompt template, an LLM draft, mandatory human editing with fact-checking, and before/after measurement (sessions, keywords in the top 10, time per article). That's proof. The certificate is only the keyword that gets you into the room where you show the proof. How to build that proof is in the article on the first marketing job.

Verdict, and what to do this week

Take a free GenAI certification now, this week, because it's immediate return and opens the ATS. The Microsoft/LinkedIn Career Essentials is a good first step, because it lands straight on your profile. Keep AI-900 for when you want exam weight. And don't stop at the badge: in the same time it takes you to stack a third certificate, build a real workflow and document it. One proves you sat down. The other proves you can do the work.

The market isn't hiring people who did an AI course. It's hiring people who solve, with it, a problem the employer doesn't even know it has yet.

This is one vertex of a larger triangle. The full map, of what a certification is worth and what it has to go hand in hand with, is in the digital marketing certifications that are worth it.

Sources

• EU AI Act - Article 4 (AI literacy)
• OpenAI launches certification program (TechRepublic)
• Google AI Essentials
• Career Essentials in Generative AI (Microsoft/LinkedIn)
• Microsoft AI-900 (Azure AI Fundamentals)

Frequently asked questions