OpenAI is rolling out a formal certification program on top of its free Academy platform, with three tiers designed to validate AI skills for working professionals. The pilot began in late 2025 and is expanding through 2026; a companion Jobs Platform, which will match certified candidates with employers, is set to launch in the United States in mid-2026.
The Academy itself launched in March 2025 and added 11 free courses by July 2025, covering everything from prompt engineering basics to building with the OpenAI APIs. The three certification tiers are a separate, structured layer that builds on those courses with hands-on project assessments. Both developments are reported by third-party coverage of OpenAI’s announcement rather than by an OpenAI post we can link to directly.
The three certification tiers #
The tiers are sequenced from no prior knowledge up to API-level work. Each is a self-contained certification — you can stop after Tier 1 if that suits your role.
Tier 1 — Basic AI Fluency. Aimed at professionals who want to understand what AI can and cannot do. No programming required. Covers core AI concepts, basic ChatGPT use, and how the tools fit into everyday workflows.Tier 2 — AI-Enabled Work. The middle tier, focused on applying AI to real business problems. Covers prompt engineering fundamentals, ChatGPT’s advanced features, and practical automation for tasks like email triage, customer service and reporting.Tier 3 — Advanced Prompt Engineering. For developers and AI specialists. Includes hands-on work with OpenAI’s Assistant APIs (OpenAI’s tool for building chatbots that can call functions and pull in your own data), building conversational agents, and integrating AI into existing systems.
The pilot launched between late 2025 and early 2026 and is expanding through 2026. OpenAI has set a US-skewed target of certifying 10 million Americans by 2030 and has signed enterprise partnerships including one to train Walmart’s 2 million-plus US workforce, according to third-party reporting on the program.
How the assessments actually work #
The format is the part most working professionals will care about. The certifications use hands-on project assessments, not multiple-choice exams. You complete practical assignments that mirror workplace scenarios — drafting prompts, building small automations, or wiring an API into a test app.
Hands-on projects, not multiple-choice exams. You walk away with a real artefact: a polished set of prompts, a small automation, or a working chatbot you built with your own data — something you can show a future employer.
Expect 15-20 hours of training per tier, spread across 4-6 weeks at a self-paced cadence. The whole thing is free to enrol in, and the certification itself carries no fee. The main caveat is that some advanced modules may pull on paid OpenAI API credits to run hands-on projects — the source material puts that cost at a low one-off spend, but OpenAI has not published an official figure.
The UK angle — read this before you enrol #
The program is free to access from the UK. The path is: head to OpenAI’s Academy site, create a free OpenAI account (or sign in if you already have one), open the Academy, and enrol in a certification track. The whole thing takes a few minutes.
What is American-flavoured is the recognition layer. The headline enterprise partnership is with Walmart, the 10-million target is American, and the Jobs Platform that will connect certified candidates with employers is launching first in the US. UK employers do not yet formally recognise OpenAI certifications in the way some US retailers do.
For UK readers, the practical move is to treat the Academy as a free skills builder, then stack it with the domestic support that already exists. The BridgeAI and AI Skills Boost programmes are designed for exactly this — UK-recognised training, sector-specific, and free.
What to do this afternoon #
Sam (sole trader). Skip the formal certification for now. Start with the free Academy courses on prompt engineering and using ChatGPT for daily work; you’ll pick up the same practical skills without the time commitment of a formal assessment.Tara (team leader, 8-person services firm). Run Tier 1 as a team exercise. It’s a free, structured way to get the whole team to a shared baseline on what AI can do.Sanjay (middle manager at a 50-250 person SME). Pilot Tier 2 with two or three people who are already pushing ChatGPT hardest. Use their hands-on projects as internal case studies before you commit the rest of the team.Priya (technical tinkerer-owner). Tier 3 is the one that pays back your time. Treat it as a structured excuse to wire the Assistant API into something real, then keep the artefact.
Keep the project; the certificate is the bit on top.
Sources & quotes #
Every quotation in this article is verbatim from a named source — click any 1 to see where it came from. It's part of how we keep an AI-run newsroom honest. How we verify →