Policy Backgrounders

The EU AI Act and the Shifting Global AI Policy Landscape

February 14, 2025

Implementation of the EU’s AI Act has begun, a significant step for the world’s first comprehensive AI regulatory regime. However, the development also highlights growing differences between the US and some allies regarding AI policy, particularly on the priority given to AI safety measures, and concerns among some policymakers and developers about the effects of regulation on innovation.

• Provisions in the EU AI Act, notably those prohibiting certain “unacceptable risk” AI systems, took effect this month. Regulators have also made progress on a draft Code of Practice governing general purpose AI models like ChatGPT. Additional provisions will take effect over the course of the next two years.
• Implementation of the AI Act highlights a growing divergence between the US and some allies regarding AI policy. For example, the new Administration, joined by the UK, declined to sign a joint declaration on AI safety developed at a recent summit and has rescinded AI policies enacted by the previous Administration focusing on AI safety and data privacy, citing concerns that the measures would stifle innovation. EU Member States and other participating countries, however, signed the declaration.
• AI firms operating across jurisdictions may face challenges navigating differing policies; in the US, action on AI legislation could shift to the states if Congress does not act.