The European Commission has challenged Anthropic's decision to limit access to its advanced AI model, Mythos 5.1, to selected US organizations. The model, priced at $10 per million input tokens and $50 per million output tokens, is designed for cybersecurity and biological research.
TL;DR
- Anthropic's Mythos 5.1 model is restricted to US organizations, raising concerns from the European Commission.
- The EU sees this as a potential economic and national security issue, as delayed access could impact European competitiveness.
- Anthropic has not provided a timetable for expanding access to other jurisdictions.
What happened
Anthropic launched Claude Mythos 5.1 on September 1, 2025, for advanced work in cybersecurity and biological research. The company describes it as its most capable model in both fields, but access is initially confined to a limited group of vetted US organizations.
The European Commission has challenged this decision, with spokesperson Thomas Regnier stating that Europe is not a security risk but an economic opportunity. The Commission is in contact with companies and administrations but has not announced formal regulatory action.
Anthropic plans to expand access but has not given a timetable or named the jurisdictions that may become eligible. The company cites the need for vetting customers and limiting deployment to reduce risk during the initial release.
Why it matters
The EU's concern is that nationality-based boundaries can look less like safety controls and more like industrial advantages. Europe has capable institutions that could meet strict access conditions, and delayed access could affect investment, talent, and operational readiness.
Access to frontier models is becoming a component of economic and national security. If European organizations receive advanced capabilities months after American competitors, it could impact their competitiveness.
The EU is trying to prepare for such restrictions with a cyber-security blueprint that envisages structured access to advanced AI capabilities and the development of sovereign European capacity.
Key facts
- Anthropic's Mythos 5.1 model was launched on September 1, 2025.
- Access is initially confined to a limited group of vetted US organizations.
- The model is priced at $10 per million input tokens and $50 per million output tokens.
- The European Commission has challenged the decision but has not announced formal regulatory action.
- Anthropic plans to expand access but has not provided a timetable or named eligible jurisdictions.
- The EU sees this as a potential economic and national security issue.
- The EU is developing a cyber-security blueprint to prepare for such restrictions.
- The case may remain a commercial dispute if Anthropic expands availability quickly.
Context
This challenge from the European Commission is part of a broader debate about Europe's dependence on American frontier AI. The EU is trying to reduce this dependence by developing sovereign European capacity and preparing for restrictions on access to advanced AI capabilities.
The case of Mythos 5.1 is a specific example of this debate. A model with potential public-security value is available in one allied jurisdiction but not another, raising concerns about economic and national security.
The EU's response to Mythos 5.1 is an early signal that it no longer regards delayed access as a routine product decision. If US-only releases become a pattern across frontier models, the Commission may use procurement, research funding, and regulation to reduce exposure.
