EU and U.S. take different paths on AI rules

The EU AI Act, spanning 113 articles and roughly 50,000 words, creates a comprehensive legal framework for AI, while the U.S. has no federal AI law and instead relies on executive orders, agency guidance and state legislation. In 2024, U.S. states passed 131 AI-related laws, more than double the 49 enacted in 2023, but EU rules can carry greater legal weight because they supersede member-state laws.

The two systems diverge sharply on prohibitions. Article 5 of the EU AI Act bans eight categories of AI practices, including social scoring, certain real-time facial recognition uses, workplace and school emotion recognition, and profiling-based criminal prediction, though four categories include exceptions. U.S. enforcement has focused more narrowly on specific applications, such as tenant screening and housing advertising, rather than banning entire technology categories.

Compliance costs are also higher in Europe. In the U.S., the annual compliance burden is estimated to be between $50,000 and $150,000, while small to medium-size enterprises in Europe face costs ranging from €50,000 to €500,000 before launch. The EU is now pursuing 10 simplification proposals targeting around €15 billion in annual savings, aiming to preserve the framework while reducing obstacles to commercial growth.