EU and U.S. take divergent paths in regulating AI

The EU AI Act, spanning 113 articles and around 50,000 words, is the first comprehensive global legislation on artificial intelligence. In contrast, the U.S. has not enacted federal AI laws, relying instead on executive orders, agency guidance, and state-level legislation. In 2024 alone, U.S. states passed 131 AI-related laws, more than double the 49 enacted in 2023. While the U.S. appears more active at first glance, the weight of its laws differs significantly. A single EU article can carry more legal force than multiple state statutes, as EU law supersedes national laws. The EU AI Act’s Article 5 bans eight AI practices, including social scoring, real-time facial recognition in public spaces for law enforcement, emotion recognition in workplaces and schools, and criminal prediction based on profiling. However, four of these categories have exceptions, such as allowing facial recognition in emergency cases. The U.S. federal government has not banned any AI technology categories, instead focusing on specific uses like tenant screening and housing advertising. Copyright laws also play a role in AI regulation, with the U.S. using the four-factor Fair Use doctrine, while the EU relies on a closed list of exceptions under the 2019 DSM Directive. The EU AI Act allows commercial AI training on data only if rights holders have not blocked it, requiring developers to audit training data against a list of opt-outs. Economists measure regulation by its impact on product costs. In the EU, high-risk AI systems, such as those used in hiring, credit scoring, and healthcare, face higher compliance costs, estimated to be two to three times more than in the U.S. Small to medium-sized enterprises face compliance costs ranging from €50,000 to €500,000, before the product is even launched. This has contributed to the relocation of nearly 30% of European unicorns between 2008 and 2021, and today U.S.-based AI companies attract 75% of global venture capital, compared to just 6% for EU companies. Oxylabs, a European unicorn that recently received a $130 million investment, has no plans to relocate, but acknowledges the challenges. EU policymakers are now calling for regulatory simplification, with the Commission proposing 10 simplification measures aimed at saving around €15 billion annually. Commissioner Dombrovskis acknowledged that excessive regulations raise costs, deter investment, and hinder innovation. The EU produces 22% of global AI research, but struggles to translate this into commercial leadership. The change in EU AI regulation is already underway, with efforts to keep the basic framework intact while reducing the burden on commercial growth.