What is GPAI?

GPAI stands for General Purpose AI — AI built for general use. These are the large, versatile AI models not designed for a single task, but usable for many different things: writing text, answering questions, summarising, translating, and much more. Well-known examples include models like GPT and Claude.

You can compare a GPAI to a broadly educated colleague who can help with almost anything, unlike a specialist who only does one thing. It is the same foundation model behind many different AI services you encounter.

Because these models are so powerful and widely used, the EU's AI Act has given them their own rules. GPAI models have specific requirements for transparency and documentation, so it should be possible to understand how they work and were trained.

Why is GPAI important for your business?

Many of the AI tools you use day to day are built on a GPAI model at their core. Understanding that helps you realise you are often building on someone else's large model, rather than creating AI from scratch.

It also affects questions of responsibility. Some requirements fall on whoever built the foundation model, while you as the user are responsible for how you apply it. Knowing where the line falls makes it easier to use AI safely and in compliance with the rules.

GPAI in practice

Say a company builds a customer service chatbot. Instead of training its own AI from zero, it uses a ready-made GPAI model and connects it to its own documents. The model provides the general language ability, and the company contributes its specific knowledge.

In that setup, certain transparency and documentation requirements fall on whoever created the foundation model, while the company is responsible for how it is used with customers — for example, being clear that an AI is answering. Understanding that split makes it easier to do the right thing.