Access to Anthropic AI models blocked: what does it mean?
Citing national security risks, the US government is forcing AI developer Anthropic to block access to its most powerful models, Mythos 5 and Fable 5, for all non-US users. In response, the company has temporarily deactivated the two models entirely. Europe's media discuss the implications of Washington's decision in the debate about digital sovereignty.
AI turned into a weapon
For the first time AI is playing a strategic role in security policy, notes De Standaard:
“The decision is driven by arbitrariness and self-interest. ... The ban prioritises US interests over public interests. And it turns AI technology into a geopolitical weapon, just as microchips were in the past. This is a sobering realisation for Europe. Although there is legislation here designed to ensure that the power of US technology companies is subordinated to the interests of consumers and free competition, it will be of little use if European companies cannot offer an alternative to Mythos, Claude or GPT 5.5.”
Not a moment to lose
Kauppalehti calls for swift action:
“While a single AI language model is not yet an irreplaceable loss for Europe, this case highlights a vulnerability in the area of digital services about which we have been warned for some time. ... People have finally woken up to the situation. In early June, the European Commission published a proposal that would restrict the use of non-EU cloud services in public sector operations. At the same time, the Commission put forward proposals to increase production capacity for AI chips and data centre capacity within the EU. ... There is no time to waste. Europe has a lot of catching up to do and advances in language models are taking place at breakneck speed.”
Optimal timing
From a European perspective the decision does at least have one advantage, the taz puts in:
“The timing is perfect. After all, which company or public authority in Europe could, in good conscience, opt for a US-based AI model right now? Especially when it's unclear whether the service will even still be available tomorrow or next week? That said, AI models are rarely – as yet – critical to the functioning of a company, organisation or public administration. The situation is quite different when it comes to operating systems or cloud services, and a similar trend is also emerging for AI. So it's time to take precautions – rather than hoping that this ban will be the last.”
Outrage is not enough
Le Monde calls for clear positioning:
“The direction is right but the pace of implementation is too slow and the scope of the measures still too limited. ... France's politicians are calling for a 'European awakening' without, however, specifying the details or the resources to be deployed. ... The presidential election campaign offers an opportunity to make progress on this front. The candidates should set out in concrete terms what they would do: what resources would they allocate to digital sovereignty, what investments they would make in European models. ... What strategy they would pursue to diversify critical infrastructure, what their stance is vis-à-vis the US. … The Anthropic affair must become the decisive argument for political action aimed at finally addressing the challenges of sovereignty – in France and across Europe.”