The Brussels effect and regulatory arbitrage
The “Brussels effect” describes the process of EU regulation influencing regulation internationally. This term is often held in a positive regard, with GDPR commonly cited as the flagship example. The EU hopes that the EU AI Act will have a similar impact. This act was the first comprehensive AI legislation passed into law, with the principal aim of balancing technological innovation and risk. However, could this hoped first-mover advantage actually become a first-mover disadvantage?
The situation
AI regulation in some form is necessary, but the EU AI Act is ~150 pages long, full of ambiguity, imposes severe penalties for non-compliance, and is yet another EU regulation to add to an increasingly long list. The Act took years to develop and will take years to fully implement, adding to the confusion. AI moves fast; policies move slow. This isn’t an environment conducive to innovation, despite what policymakers may claim.
Why this matters
In response, startups and talent might seek opportunities for regulatory arbitrage and base themselves in nations with lighter regulations. If those nations see an increase in AI startups, inbound talent and investment, why would they follow the EU’s lead and risk a competitive edge by introducing similarly restrictive regulation?
Recent research seems to support this view, highlighting a significant gap in AI adoption between the UK and the EU due to regulatory concerns, with Jacob Beswick, Senior Director of AI governance at Daikatu saying: “The EU AI Act has raised more questions than answered, and in the process, businesses within its jurisdiction have become increasingly hesitant about their AI programs”.
There are counter arguments, of course. The EU AI Act is risk-based, meaning there are varying levels of compliance mandates dependent on the risk of a given AI system, potentially reducing the burden on firms. The EU is obviously a large market which is another factor; firms wanting access will have to comply. Diverging too far from the EU’s regulatory approach might increase complexity, adding friction and hampering innovation and global interoperability. And if an AI-related safety incident occurs in a country with fewer regulations, public pressure may force regulation, and the EU may be proven right.
The EU AI Act will influence global AI regulation, that much is true. But will it be as a success story or a tale of caution?
References / Related reading
How Europe Became the World's Top Tech Regulator (CNBC)
Raise the Bar: Leveraging the EU’s Regulatory Power (Chatham House)
EU AI Act: First Regulation on Artificial Intelligence (European Parliament)