The EU has already introduced four initiatives to reduce exposure to US technology policy shifts.
Following the Trump administration’s order on 12 June 2026 requiring an AI giant to cut off foreign nationals from accessing its top-tier AI models, the European Commission’s Tech Sovereignty Package, released on 3 June 2026, has gained heightened significance regarding the bloc’s exposure to US technology policy shifts.
The tech sovereignty package brings together four key initiatives: Chips Act 2.0, the Cloud and AI Development Act known as CADA (a four-level sovereignty classification system for cloud and AI infrastructure), an EU Open Source Strategy, and a Strategic Roadmap for Digitalisation and AI in Energy. The Commission describes this collection as a “major shift” in technology policy designed to cut structural reliance on non-European suppliers.
The proposal mandates that member nations evaluate their cloud providers’ sovereignty risks and match procurement to those findings, with the most sensitive government systems confined to EU-controlled services. Level 4 represents the strictest requirement, demanding complete EU ownership and control, personnel with EU clearance, zero transfer of AI inference data beyond European borders, and third-party audits certified by national regulators. The legislation also seeks to multiply EU data center capacity by three within five to seven years.
Chips Act 2.0 pursues €120 billion in investment through 2035, with a goal of manufacturing at least 20% of global cutting-edge semiconductors by 2030. Both proposals move forward to the European Parliament and Council for negotiation, expected to take 18 to 24 months.
Ban confirms European worries
According to Reuters, Anthropic had declared it was shutting down its Fable 5 and Mythos 5 models for everyone after receiving a US Commerce Department export control order prohibiting foreign nationals from using them. The firm stated it had received no specific explanation about the national security issue.
The shutdown has demonstrated exactly the dependency threat that the Commission’s package targets: European governments and businesses depending on US AI vendors risk losing access instantly based on Washington decisions. Also, Anthropic’s forced exit occurring just 10 days after the Commission’s proposals bolstered EU officials arguing the bloc required binding sovereignty rules instead of voluntary pledges.
A Commission official had told Reuters the proposals seek to ensure Europe “can develop, deploy and secure” technologies meeting its own requirements. The tech sovereignty package commits €2 billion across seven years to an open-source strategy intended to expand European alternatives in cloud computing, AI, and cybersecurity. A “Free Software first” rule would govern public cloud and AI procurement, aiming for 30m active users of open-source collaboration platforms by 2030.
