By Martin Coulter, Supantha Mukherjee and Foo Yun Chee
LONDON (Reuters) -The European Union’s landmark Artificial Intelligence (AI) Act may exempt open-source models from strict regulation, according to a leaked compromise proposal seen by Reuters.
Lawmakers halted negotiations on Thursday, after almost 24 hours of overnight negotiations failed to reach a consensus on how best to regulate the rapidly-developing technology.
Open-source refers to the free and open sharing of software code, allowing anyone to contribute to upgrading it or resolving bugs.
Notable European companies operating in the open-source space include France’s Mistral and Germany’s Aleph Alpha, both of which had previously criticised European proposals to regulate the technology.
According to the document, which circulated among lawmakers on Thursday morning, the AI Act would not apply to free and open-source licences unless, for example, they were deemed high-risk or being used for already banned purposes.
France, Germany, and Italy had previously pushed for makers of generative AI models to be allowed to self-regulate, in an apparent effort to support local startups.
An exemption for the open-source models built by Mistral and Aleph Alpha could be seen as a victory on their part, according to one source familiar with negotiations. “The law isn’t settled yet, however, so we can’t rush to conclusions,” they added.
Microsoft-backed OpenAI was also founded as an open-source nonprofit, before co-founder Sam Altman pivoted to a capped-profit structure in 2019.
(Reporting by Martin Coulter, Supantha Mukherjee and Foo Yun Chee; Editing by Jan Harvey, Mark Potter and Emelia Sithole-Matarise)
