EU’s Artificial Intelligence Safety Bill Reaches Final Stages

The EU’s artificial intelligence (AI) safety bill is set to enter its final stages of the legislative process, known as “trilogues,” where the EU parliament, commission, and council will determine the final content of the bill. The bill aims to regulate AI technology based on its potential to cause harm. However, internal disagreement regarding the regulation of “foundation” AI models trained on large datasets poses a challenge to the bill’s progress. These models, known as general-purpose AI (GPAI) systems, can perform a variety of tasks and are expensive to develop. Only a select few companies can afford to invest in this technology.

Key Points:

• The EU’s AI safety bill is the first attempt to seriously regulate AI technology based on its potential to cause harm.
• The bill is now in the final stages of the legislative process, where key disagreements regarding the regulation of GPAI systems may hinder its progress.
• GPAI systems are expensive to develop, and only around 20 global firms have the financial resources to invest in this technology.

Despite the high stakes and the importance of this bill in shaping AI regulation, internal disagreements concerning the regulation of GPAI systems threaten its progress. With only a handful of companies able to afford the cost of developing these systems, it remains to be seen how the proposed legislation will address the challenges and risks associated with AI technology. As the trilogues begin, all eyes will be on Brussels to see how this landmark bill evolves.