AI sovereignty is no longer a theoretical issue

Over the last few hours we have seen a very concrete reminder of something that is often treated as a theoretical debate: AI sovereignty is not an abstract topic anymore. It is an operational and strategic one.

Anthropic suspended model access for Fable5 and Mythos5, citing national security concerns and possible use by foreign actors. Whatever position one takes on the specific case, the broader lesson is clear: if your AI capability depends entirely on an external provider, you are also accepting that provider's constraints, policies and unilateral decisions.

What AI sovereignty actually means

AI sovereignty is not only about where a model runs. It is about retaining meaningful control over:

• access to models and service continuity
• data, logs, telemetry and compliance boundaries
• the ability to switch providers without freezing the business
• technical, contractual and geopolitical governance

Why this matters for Europe, public sector and enterprise

For Europe, this is especially important. If infrastructure, models and decisive policy layers all remain outside our perimeter, then our autonomy is fragile even when the solution looks excellent in a demo.

The same logic applies to public administration, healthcare, regulated industry and large enterprises. Buying AI capability is not enough. We also need to understand which dependencies we are introducing and how much real control stays in-house.

The real strategic question

The question is not only: "Which model should we choose today?"

The more important question is: "How much control will we still have tomorrow if access, pricing, regulation or geopolitical conditions change?"

A pragmatic direction

That is why I believe the topic needs action on multiple levels:

• multi-provider architectures, or at least provider-switchable ones
• explicit governance for critical data and workloads
• geopolitical risk assessment alongside technical risk assessment
• European investment in infrastructure, models and strategic supply chains

AI sovereignty does not mean isolation. It means having the ability to choose, negotiate and continue operating when the environment changes.

That is the moment when the topic stops being theoretical and becomes architecture, industrial strategy and technology policy.