By Holt Hackney
The European Union has begun enforcing additional provisions of its Artificial Intelligence Act, expanding regulators’ authority over AI developers and systems, including companies based outside Europe.
The latest phase of the law took effect Aug. 2, giving the European Commission’s AI Office and national authorities greater enforcement authority over provisions involving general-purpose AI models. The broader law also establishes transparency requirements for certain AI systems, including rules governing chatbots, deepfakes and AI-generated or manipulated content.
The development has renewed debate in the United States over the extent to which European regulations could influence how American technology companies develop and deploy artificial intelligence.
The AI Act applies not only to companies located within the European Union but also, under certain circumstances, to providers and deployers outside the bloc when their AI systems are offered in the European market or their outputs are used within the EU.
That provision potentially places U.S.-based AI companies under European regulatory requirements even when their technologies are developed primarily in the United States.
The Effect on American Technology Policy
The Heartland Institute, a U.S.-based conservative and libertarian public policy organization, has argued that the law’s international reach could affect American technology policy.
“The EU AI Act is no longer simply a regulatory proposal whose international effects can be debated in the abstract,” said Jack McPherrin, a senior policy analyst and research fellow at Heartland. “European authorities now have meaningful enforcement powers over rules that reach American companies and technologies.”
Heartland examined the issue in a 2025 policy study, “The European Union’s Artificial Intelligence Act: An Extraterritorial Incursion on Americans’ Inalienable Rights.” The organization argued that global AI companies could have incentives to apply European standards more broadly rather than maintain substantially different versions of their products for different markets.
The EU law takes a risk-based approach to AI regulation. It prohibits certain uses considered unacceptable while imposing varying obligations on other systems depending on their potential risks. The law also establishes requirements involving transparency, copyright, model safety and systemic risks associated with some general-purpose AI models.
Some provisions affecting high-risk AI systems are scheduled to take effect during later stages of the law’s implementation.
Supporters of the EU approach have argued that common standards are needed to address potential harms from rapidly developing AI technologies and provide greater transparency and protection for consumers. Critics have raised concerns about regulatory costs, restrictions on innovation and the international influence of European rules.
Donald Kendal, director of Heartland’s Glenn C. Haskins Emerging Issues Center, said the United States should develop its own approach rather than allow European standards to become the de facto framework for American companies.
“The United States should reject this attempt to export Europe’s regulatory regime and instead pursue policies that protect free speech, individual liberty, technological innovation, and American sovereignty,” Kendal said.
The debate comes as governments worldwide consider how aggressively to regulate generative AI and other rapidly developing technologies. The EU’s enforcement framework could become particularly significant because major American AI developers seeking access to European consumers must comply with applicable provisions of the law.
McPherrin said U.S. policymakers should pay particular attention to that cross-border effect.
“U.S. policymakers should take that extraterritorial reach seriously and ensure that decisions governing American AI remain accountable to Americans,” he said.