General-purpose models have their own chapter
GPAI models are regulated separately from AI systems. Every GPAI provider must keep technical documentation, give downstream providers what they need to comply, put a copyright policy in place, and publish a sufficiently detailed summary of the training content.
Why it is like this
A model that can be put to any purpose cannot be classified by use. The chapter regulates it at the layer where it exists — the model — and leaves the use to whoever deploys it.
Not to be confused with
A high-risk classification. A GPAI model is not high risk by virtue of being general; the obligations are documentation and transparency, and the risk tier attaches to the systems built on it.
In practice
The training-content summary follows a template from the AI Office — it is a public document, not an internal one.
Source: Regulation (EU) 2024/1689 (AI Act) — Article 53. Read on .
Say it back Check yourself
Name the four baseline obligations on a provider of a general-purpose AI model.
Answer out loud or in writing first, then open the points below.
The points a good answer makes
- Draw up and keep technical documentation of the model
- Provide information and documentation to downstream providers integrating it
- Put in place a policy to comply with EU copyright law, including the text-and-data-mining opt-out
- Publish a sufficiently detailed summary of the content used for training
Open-source models under a free licence are exempt from the first two — but never from the copyright policy or the training-content summary.