European Commission Publishes Training Data Summary Template for Providers of GPAI Models

Data training transparency is a core obligation under the AI Act (Regulation (EU) 2024/1689) for General-Purpose AI (GPAI) models placed on the Union market, including those distributed under free and open-source licenses. To support compliance with Article 53 of the AI Act, the European Commission has introduced a standardized Template for the Summary of Training Content — a practical tool designed to facilitate consistent, rights-respecting disclosures by model providers. This obligation applies as of 2 August 2025 for GPAI models placed on the Union market from that date onward, while providers of models already on the market before that date must comply by 2 August 2027.

The Template consists of three main sections:

General Information
This section captures basic details about the model and its provider, as well as general characteristics of the training data. It aims to establish a baseline understanding of the dataset’s scope and relevance.

List of Data Sources
Model providers are required to enumerate the data sources used, including both public and private datasets. While it does not mandate disclosure the details for the specific data, the template calls for a narrative description of dataset categories and, where applicable, the top domain names of online sources crawled or scraped.

Relevant Data Processing Aspects
Providers must articulate how they respect legal frameworks governing text and data mining (TDM), including opt-outs. Additionally, the template requires disclosure of measures taken to identify and exclude illegal content from training datasets.

Providers may nevertheless decide on a voluntary basis to go beyond the minimum requirements in the Template.

Together with the Commission’s Guidelines on the Scope of the Rules for GPAI Models and the GPAI Code of Practice, this template forms part of a AI governance framework. It is intended not merely as a compliance aid, but as an instrument to foster transparency, legal certainty, and public trust throughout the AI value chain.

The Template and more information could be found here.

See All News Here

Related News

Swedish Supreme Court Adds Nuance to Previous Ruling on Bulk Requests for Court Judgments and Decisions

NIS2 and the Swedish Regulation Jungle

EU Commission Proposes Tech Sovereignty Package with New Rules for Chips, Cloud Services and AI infrastructure