The European Union has raised concerns with the Trump administration and Anthropic over the US startup’s move to cut foreign access to its Fable 5 and Mythos 5 AI models.

European Commission Executive Vice President Henna Virkkunen said the change blocked European cybersecurity use.

Anthropic said the restriction followed findings that users could “jailbreak” Fable 5’s safeguards. Virkkunen said Washington tied the curbs to national security concerns..

The block also affects official EU work.

Anthropic had approved the European Union Agency for Cybersecurity in early June to test Mythos through Project Glasswing, a restricted program for evaluating the vulnerability-finding model before a wider release.

The European Commission plans an AI and cybersecurity action plan in early July, ahead of August AI Act milestones, as Brussels clarifies how rules for general-purpose AI models will apply.

🔗 Source: Bloomberg

Recent Anthropic developments