European lawmakers are facing continued disagreements on various issues related to new artificial intelligence (AI) rules ahead of a crucial meeting set for Tuesday. According to sources familiar with the matter, any agreement will now be postponed until December.
The draft AI regulations must be reached through an agreement between the European Parliament and the European Union member states. These regulations have been discussed three times in trilogies, which are meetings aimed at finalizing laws by bringing together parliament and EU states, per Reuters.
A fourth trilogue meeting is scheduled for Tuesday, following discussions by EU lawmakers about their negotiating stance regarding foundation models and high-risk AI systems.
Foundation models, such as OpenAI’s ChatGPT, are AI systems that are trained on large datasets and can learn from new information to perform various tasks.
Spain, which currently holds the EU presidency until December, has been actively pushing for a deal and has proposed compromises to expedite the process. These compromises include a tiered approach for regulating foundation models, specifically those with over 45 million users, as outlined in a draft seen by Reuters.
Spain is also advocating for additional obligations concerning very capable foundation models (VCFM), including regular vetting to identify potential vulnerabilities. Opponents argue that smaller platforms can also pose significant risks.
Spain has consulted with other EU countries regarding potential compromises ahead of the upcoming fourth trilogue. However, it is unlikely that a final agreement will be reached during this meeting, as indicated by the sources.