EU/US transatlantic AI policy divergence becomes structural — EU adds nudifier prohibition Dec 2, US EO consolidates federal preemption
The EU AI Act omnibus (May 7) and the Trump December 2025 EO are now formally moving in opposite directions. The EU adds specific prohibitions enforced under GDPR-style long-arm jurisdiction; the US consolidates federal oversight to preempt state-level AI regulation. The transatlantic policy divergence is no longer a transient negotiating gap — it's the operating reality for any AI company that ships to both jurisdictions.
The structural divergence shows up in three layers. First, prohibitions: EU adds specific categories (nudifier apps effective Dec 2 2026); US has none and has signaled it will not. Second, enforcement architecture: EU uses GDPR-style fines and market-access restrictions; US uses federal preemption of state law as the primary leverage. Third, compliance cost: EU's high-risk system regime imposes substantial documentation and audit requirements; US EO directionally reduces compliance burden by consolidating oversight federally.
For multinational AI companies the consequence is operational. The 2024 expectation that EU and US frameworks would gradually converge through coordination has been replaced by the 2026 reality that they're diverging through deliberate policy choice. The compliance architecture has to be two distinct stacks, not one harmonized stack. New York's state-level AI safety law (which the December EO is intended to preempt) is the visible domestic flashpoint; the international divergence is the bigger long-term cost.
Data Innovation — New York's AI Safety Law Claims National Alignment but Delivers Fragmentation → · Consilium — AI Council and Parliament agree to simplify and streamline rules → · National Law Review — What the Regulations of 2025 Could Mean for the AI of 2026 →