Repository logo
 

Open-source provisions for large models in the AI Act

Published version
Peer-reviewed

Type

Article

Change log

Authors

Law, Harry 
Krier, Sébastien 

Abstract

On 21 April 2021, the European Commission tabled a proposal for the European Union’s Artificial Intelligence Act to introduce a common regulatory and legal framework for the development and deployment of artificial intelligence (AI). Subsequent amendments have sought to include generalist or ‘foundation’ models, including in some commercial contexts those released on an open-source basis. Critics have argued that such a move would harm smaller AI developers, who rely on open-source practices to develop new products and services. Others contend that targeted measures are necessary, given the risks of misusing open-source systems. This paper focuses on approaches to open-sourcing foundation models in the context of the Act, rather than questions relating to general open-source practices or publication norms for AI systems. It seeks to summarise the movements related to open-source models in the Artificial Intelligence Act and introduces possible avenues for compromise.

Description

Keywords

Artificial Intelligence Act, AI Act, general purpose AI systems, foundation models, GPAIS, open-source, artificial intelligence, AI, machine learning

Journal Title

Cambridge Journal of Science and Policy

Conference Name

Journal ISSN

Volume Title

4

Publisher

Cambridge University Science and Policy Exchange

Publisher DOI

Publisher URL