US controls on frontier AI expose Europe’s dependence problem

🔍 WHY THIS STORY MATTERS
Access to frontier AI is becoming a security decision, not just a commercial release. Europe’s dependence on US models could affect cyber defence, public services, industry and democratic oversight.

KEY TAKEAWAYS...

● US interventions affect access to advanced Anthropic and OpenAI models.

● The restrictions are linked to cyber-security concerns around powerful AI systems.

● European users remain dependent on US companies, infrastructure and political decisions.

● AI sovereignty will require capability, bargaining power and transparent allied rules.

E urope’s debate over artificial intelligence often centres on regulation, ethics and industrial ambition. Recent restrictions on advanced models from Anthropic and OpenAI have exposed a harder question: what happens when access to powerful AI systems used by European companies, governments and cyber-security teams can be shaped by decisions made in Washington?

The issue became urgent in June, when Anthropic took its Fable 5 and Mythos 5 models offline after a Trump administration directive blocked their use by foreign nationals. The company said it had disabled the models to comply with the order. The stated concern was national security, with particular attention to the ability of advanced systems to help identify and exploit software vulnerabilities.

Anthropic later restored wider access to Fable 5, but the more capable Mythos 5 remained restricted to a smaller group of users. The dispute did not stay inside one company. A group of cybersecurity executives and experts urged the White House to ease the Anthropic restrictions, arguing that blocking trusted foreign defenders could weaken, rather than strengthen, collective security.