According to "Business Insider", the United States has implemented export controls on Anthropic's latest AI model, which is exactly what its opponents want. Mistral AI, Europe's number one AI company and known as the "European version of OpenAI," has been warning of the risks of relying on American AI companies.
Mistral CEO Mensch has been warning about the risks of relying on the U.S.
On Friday, U.S. officials imposed export controls on Anthropic's cybersecurity-focused models Mythos 5 and Fable5. Citing national security concerns, they said security safeguards designed to prevent Fable 5 from being abused could be bypassed.
In response, Anthropic suspended access to these models entirely, raising uncertainty about who ultimately controls access to cutting-edge AI.
The incident fits exactly what Mistral CEO Arthur Mensch has been warning about for more than a year. He later used this as part of his promotional strategy to explain why people should choose their model instead of models developed by American companies such as OpenAI and Anthropic.
In a speech at London Technology Week in June 2025, Mensch warned that American AI companies held the "keys" to their models, which was tantamount to European companies "giving the initiative to suppliers."
"At some point you need to have the ability to turn these systems on or off and not give that power to another country," he said.
Since then, Mistral has based its strategy on "AI sovereignty," arguing that governments and companies should not be overly dependent on a handful of foreign AI providers. To this end, it has been promoting "open weight models", allowing customers to deploy these models on their own infrastructure and use their own data to make customized adjustments.
"European governments come to us because they want to build their own technology and provide services to their own citizens." Mensch said in an interview with the "Big Tech Podcast" in January this year.
Last month, Mensch further strengthened his stance at a hearing in the French National Assembly on digital sovereignty and AI. He warned that Europe only has two years to build its own AI infrastructure, otherwise it will be permanently dependent on American technology giants.
While Mistral is generally considered the leading AI model provider in Europe, it still lags behind the likes of Anthropic in terms of valuation, model capabilities and user scale. Mistral was last valued at about $13.6 billion, while Anthropic was valued at about $965 billion.
However, the restrictions imposed by the United States on Anthropic make Mistral’s core argument easier to understand: in most cases, control of AI will ultimately rest with the provider and the government of the country to which it belongs.