The White House’s move to restrict access to Anthropic’s latest AI models could give Europe’s leading AI startup exactly the opportunity it has been preparing for.
On Friday, U.S. authorities imposed export restrictions on Anthropic’s cybersecurity-focused models, Mythos 5 and Fable 5. They cited national security concerns that safeguards meant to prevent Fable 5 from being exploited could be circumvented.
This restriction prevents foreigners from accessing the two models. Anthropic responded by completely shutting down access to its systems, creating uncertainty as to who would ultimately control access to Frontier AI.
The episode bears the hallmarks of a scenario Mistral CEO Arthur Mensch has been warning about for more than a year, and has since become part of the pitch for why people should choose Mistral over models from U.S. companies like OpenAI and Anthropic.
Speaking at London Tech Week in June 2025, Mensch warned that US AI companies “hold the keys” to their models, adding that he sees European companies “giving leverage to providers.”
“At some point, we need to be able to turn it off and on, and we don’t want to leave that to other countries,” he says.
Sovereign AI
Since then, the French startup has positioned itself around AI sovereignty, which says governments and businesses should not rely too heavily on a small number of foreign AI providers.
To that end, we championed an open-weight model that customers can deploy on their own infrastructure and customize with their own data.
“European governments are coming to us because they want to build technology and they want to provide services to their people,” Mensch said on the Big Tech Podcast in January.
Mensch emphasized his position at a hearing on digital sovereignty and AI in the French National Assembly last month, warning that Europe has two years to build its own artificial intelligence infrastructure before becoming permanently dependent on US tech giants.
At Mistral’s first summit in Paris that same month, executives, government officials, and business customers repeatedly returned to the same theme: AI sovereignty.
“You need to know where your data is and what’s going to happen to it,” Jan van den Bremen, Accenture’s head of technology for Europe, Middle East and Africa, told Business Insider at the event.
Although Mistral is widely considered to be the top AI model provider in Europe, it lags behind the likes of Anthropic in terms of ratings, model capabilities, and number of users. The French startup’s final valuation was around $13.6 billion, compared to Anthropic’s valuation of $965 billion.
But human limitations make Mistral’s core argument easier to understand. In most cases, control of AI ultimately lies with the provider and its government.