Study: EU's AI guardrails too rigid for technological change
New research warns that the EU's artificial intelligence regulatory framework is too rigid and ill-equipped to adapt to rapid technological advancements.
A new study published in the journal "Big Data & Society" cautions that the European Union's "guardrail" mechanism for regulating artificial intelligence (AI) suffers from deficiencies in both policy and practical implementation. Researchers argue the regulatory system is too rigid to adapt to AI's rapid evolution.
The research indicates that the EU's framework was formed through years of negotiation and political coordination, making it highly difficult to amend. Conversely, repealing certain provisions is simpler, creating a "rigidity trap." The AI Act, originally scheduled for 2025, has been replaced by the "AI Simplification Act," signaling a slower regulatory pace.
The scholars contend that the EU's current framework has failed to achieve its objective of trustworthy, human-centric AI that respects fundamental rights. In contrast, the U.S. regulatory model is described as a "tow rope," capable of tightening or loosening as needed, offering greater enforceability and faster implementation than the EU's "guardrails."
Furthermore, the EU system attempts to anticipate all potential AI risks and establish comprehensive regulations. However, the study notes that it is impossible to foresee all risks in advance, leading to implementation challenges and limited public value. The U.S., on the other hand, typically enacts regulations targeting specific companies or industries once concrete risks emerge.
The research was a collaborative effort by the University of Exeter, the European University Institute, and the University of Lausanne.