European lawmakers are facing challenges in reaching an agreement on new artificial intelligence (AI) rules, postponing any potential deal until December. The draft AI regulations require approval from both the European Parliament and EU member states, with three previous rounds of discussions in trilogues between the two entities.
A fourth trilogue meeting is scheduled, focusing on foundation models and high-risk AI systems. Spain, the current EU presidency holder, has been pushing for a deal, suggesting compromises like a tiered approach for regulating foundation models with over 45 million users and additional obligations for very capable foundation models, such as ChatGPT.
Opponents argue that smaller platforms can also pose risks. While Spain has consulted with other EU countries on potential compromises, reaching a final agreement in the upcoming meeting is unlikely. A fifth trilogue is set for early December, and failure to reach a deal could push negotiations into early next year, complicated by European Parliament elections in June. Despite some lawmakers expressing hopes for approval by the end of the year, issues related to AI rules and their implications continue to pose challenges, particularly as AI tools are to be classified based on their perceived level of risk.
The proposed classifications range from low to unacceptable risk, with corresponding obligations for governments and companies using these tools, depending on the level of risk they pose. The struggle to reach consensus highlights the complexity of regulating AI at a European level and balancing innovation with safeguards and accountability.