Nathalie Moreno: AI regulation: The EU–US sovereignty contest

AI regulation has emerged as a key legal instrument through which the European Union and the United States pursue competing objectives of technological sovereignty and technological leadership. In the EU, the AI Act now sits alongside broader regulatory measures addressing dependence on third-country providers, compute, cloud infrastructure and semiconductors. Conversely, US AI policy extends to export controls, procurement and national-security measures to project regulatory influence with far-reaching cross-border effects. This friction establishes an intense EU–US sovereignty contest, which we examine first through the lens of competing jurisdictional claims over AI deployment, and then, by analysing the emergence of EU sovereignty frameworks designed to counter US technological dominance.

1.      Competing claims of jurisdiction over AI development and deployment

We consider the extraterritorial application of EU and US AI laws as well as their competing regulatory objectives.

A.    Impact of the extraterritorial application of EU and US AI laws

The EU AI Act’s extraterritorial application extends to third-country providers, deployers, importers, and distributors when placing AI systems or general-purpose models on the EU market, putting them into service, or where the output produced is intended for use within the EU. US export-control and national-security rules may concurrently affect access to AI models, compute or controlled technology. Therefore, whilst EU law may govern market placement or use, US law governs supply.

B.     Clash between EU fundamental-rights-led regulation and US innovation-led AI governance

The EU AI Act aims to combine harmonisation and innovation with safety and protection of fundamental rights under the European Charter of Fundamental Rights. On 21 August 2026, the Dutch Data Protection Authority announced fines amounting to €824.99 million against Uber B.V. and Uber Technologies Inc. for solely automated decisions temporarily or permanently deactivating drivers’ platform accounts, in breach of Article 22 of the GDPR. US federal policy differs. Executive Order 14179 of 23 January 2025 directs review and removal of measures impeding US AI leadership and the White House legislative recommendations of 20 March 2026 call on Congress to remove unnecessary barriers and pre-empt burdensome State AI laws. On 24 April 2026, the Department of Justice moved to intervene in xAI’s challenge to Colorado’s algorithmic-discrimination law, a dispute that prompted Colorado to rewrite its regime away from the EU AI Act ex-ante rights model.

2.      The emergence of EU AI sovereignty regulation in the face of US technological leadership

We now consider the EU regulatory response to non-European AI dependence and its limits while critical capability remains subject to US law.

A.    EU technological sovereignty as a response to dependence on US AI capability

On 3 June 2026, the European Commission presented the European Technological Sovereignty Package, including proposals for the Cloud and Data Act (CADA) and Chips Act 2.0. CADA is significant because its explanatory memorandum states that the AI Act “does not cover aspects of sovereignty”. Its proposed four-level cloud sovereignty framework would permit certain third-country-controlled providers to be audited against level 3 only where the Commission recognises the relevant country by implementing act against criteria relating to governmental access under Article 32 of the Data Act, service continuity, access to advanced technology and procurement reciprocity. They directly apply to US providers subject to US law. Together with Chips Act 2.0, the package shifts EU policy away from regulating AI systems towards reducing dependencies in the infrastructure they rely on.

B.     The limits of EU AI sovereignty under US legal control of critical AI capability

However, US technological dominance is sustained by expansive federal statutory powers and unprecedented capital market concentration. Executive Order 14320 of 23 July 2025 established the American AI Exports Program to promote full-stack US-origin packages and worldwide adoption of US technologies. On 12 June 2026, Anthropic stated that a US export-control directive required it to deny foreign nationals access to Claude Fable 5 and Claude Mythos 5, forcing global suspension until the controls were lifted on 30 June 2026. As such, whilst EU law can govern AI deployment in Europe, US public law can alter the availability of the underlying AI capability.

Conclusion

The transatlantic contest combines jurisdictional overlap with asymmetric control over the infrastructure and capability on which AI deployment depends. Whether the EU’s emerging technological-sovereignty framework can secure meaningful European AI sovereignty while dependence on AI capability remains subject to US law remains the central question.