EU moves to secure AI model access and reduce US cloud dependence

The European Commission released an action plan (COM(2026) 577) that outlines nine measures to protect the EU’s access to advanced AI models. It includes contingency procedures for a scenario where a provider or a third‑country authority blocks a model, a situation triggered weeks earlier when the U.S. Commerce Department halted Anthropic’s Claude Fable 5 and Claude Mythos 5 for foreign nationals, causing a 19‑day global shutdown. An IBM and Oxford Economics study found that four‑fifths of EMEA firms would suffer severe to critical impacts from a similar outage lasting a week or more. The Commission aims to publish a European Blueprint by Q4 2026 detailing which EU organisations may use such models, without imposing binding duties on non‑EU providers.

At the same time, European software firms are pushing to lessen reliance on U.S. cloud services. On 22 July 2026, Austrian company sproof and German firm Governikus announced a strategic alliance to integrate Germany’s eID into sproof’s identity platform, preparing for eIDAS 2.0 and the European Digital Identity Wallet from 2027. Open‑source solution openDesk, after tests with the German pension and employment agencies, received approval for emergency‑workplace use. With U.S. hyperscalers holding roughly 70 % of the European cloud market—and over 90 % in German public administration—initiatives such as NEXOWARE’s Europe‑hosted AI‑agent platform and cashwerk.io’s AI‑driven e‑form builder aim to bolster digital sovereignty.