Spain, the current holder of the presidency of the Council of the European Union, is trying to rescue the bloc’s first stab at regulating artificial intelligence (AI) amid a revolt led by France.
French President Emmanuel Macron’s Government has been joined by the German and Italian authorities, all of which are concerned about the push to control so-called “foundation models”, a category including the likes of the US OpenAI-created GPT-4 and DALL-E.
While Eurocrats are keen to see stringent controls put on to such systems, France is concerned that Brussels-style bureaucracy will further undermine Europe’s efforts to catch up with China and the US in the worldwide AI race.
With the French coalition now allegedly stonewalling attempts to push negotiations on the regulation through its final stage, Spain is trying to build a last-minute compromise between the parties.
The nation has floated a compromise version of the proposed regulation that thins out some of the restrictions on foundation models.
Said to be based on a previous compromise drawn up by the European Commission, the draft document also includes new recommendations from the European Parliament, which remains adamant that strict controls are needed.
MEPs are particularly concerned about the widespread misuse of the technology in the realms of both business and governance, with some also fearing that the technology could end up discriminating against minority groups.
The chamber voted to ban authorities from using facial-recognition artificial intelligence (AI) technology, in a move described as “a historic and great success”. https://t.co/PElVPfK16q
— Brussels Signal (@brusselssignal) June 15, 2023
It seems unclear whether or not the Member State coalition will take on the new variant. The document maintains strict controls on foundation models once they breach one of two technical thresholds on users and training data.
They have instead pushed for a style of self-regulation in the sector based on an EU-wide code of conduct for companies building foundation models rather than hard restrictions based on specific pieces of technical information.
Other experts have also warned against Eurocrats’ efforts to regulate foundation models.
“The problem is not the technology,” Hosuk Lee-Makiyama, the director of European Centre for International Political Economy, told those gathered at the Techno Politics conference in Brussels earlier in November.
He noted that even the most basic resources, such as electricity provision, could be misused but added that such a possibility was not reason enough to prompt EU-level regulation.
Lee-Makiyama said there was already widespread confusion among both individuals and businesses about European tech regulations and that further restrictions on foundation models would not help matters.
“It’s not just about business – users need clarity. They need stability and predictability,” he said, adding that Brussels passing a law restricting foundation models could lead to legal “uncertainty”.
Seemingly unhappy with just ruling Europe, MEPs have demanded that the European Union starts regulating “virtual worlds”, such as Facebook’s metaverse and Star Trek-style holodecks. https://t.co/QjNcOzKqnr
— Brussels Signal (@brusselssignal) November 28, 2023
