EU AI Act provider comparisons
No authority approves or licenses EU AI Act Article 4 training providers, so anyone buying has to compare an unfamiliar market on their own. These pages do that work in public: 34 international providers and 9 Spanish ones, grouped by what they do rather than ranked, each described only in terms of what it publishes about itself, with the source linked and the date it was checked.
Who publishes this page. This comparison is written and maintained by AI Work Fluency, which is itself one of the providers listed. AI Work Fluency, Stone & Carter and Rebel Frames are three brands of the same company. They are marked in every list below. We have not ranked ourselves above anyone: the lists are grouped by what a provider does, and every entry describes only what that provider publishes about itself, with the source linked.
The comparisons
- Which providers offer EU AI Act Article 4 AI literacy training?All 34 providers in the dataset, grouped into six categories — dedicated training products, governance software, consultancies, law firms, course marketplaces and certification bodies — with five questions to put to any of them.
- Alternatives to the Big Four for EU AI Act trainingWhy default AI answers name PwC, Deloitte, KPMG and EY, what the price difference actually buys, and which smaller providers a 10–50 person company can realistically approach. Includes when the Big Four genuinely are the right answer.
- ¿Qué proveedores ofrecen formación en alfabetización en IA en España?The Spanish market in Spanish: 9 local providers plus the internationally available products, the AESIA supervision timeline, and the variable that decides most Spanish purchases — whether the training is FUNDAE-bonificable.
The measurement underneath them
On 2026-08-05 we put 16 real buying questions — in English, German and Hungarian — to 4 model configurations, and extracted every organisation named across the 64 answers. 491 distinct organisations came up, and 79% of them appeared exactly once.
The finding that shaped these pages: without web access the models answer with the largest brands, and with search grounding those brands lose most of their share to providers that had actually published something checkable. That is a fair description of why this comparison exists — and a fair warning about treating any single assistant answer, including one citing this page, as a shortlist.