France names Mistral sovereign AI provider, explicitly excluding OpenAI
On Aug 22-23 French Budget Minister David Amiel said the government will hire sovereign AI providers such as Mistral to test public services for vulnerabilities, explicitly excluding OpenAI. The move follows a DGFiP tax-agency breach exposing roughly 700K taxpayers.
On August 22-23, French Budget Minister David Amiel announced that the government will hire sovereign AI providers such as Mistral to test public services for vulnerabilities, explicitly excluding OpenAI from sensitive government AI workloads. The decision follows a recent breach at DGFiP, the French public finance directorate, which exposed the personal information of roughly 700,000 taxpayers.
France's explicit designation of Mistral as a government sovereign AI provider marks another concrete step in Europe's push for AI sovereignty on sensitive government workloads. The government's logic is that AI sovereignty requires not just local deployment capability but full control over model supply chain, training data and inference environment. This also raises the entry barrier for US-headquartered AI companies in the European government market.
Mistral had already launched an upgraded Agentic Search on August 20, positioning retrieval-augmented capability as a differentiated offering for highly regulated industries such as finance and law. The retrieval layer reports accuracy jumping from 26.7% to 86% on FinanceBench, with token consumption cut by one-third and p90 latency reduced by up to 39.6%. With the French government designation on top of its existing enterprise customer base, Mistral is becoming the representative vendor along Europe's sovereign AI path.
Following the DGFiP incident the French government decided to migrate all AI systems handling sensitive taxpayer data to certified sovereign AI providers, with Mistral and Germany's Aleph Alpha seen as the first batch of candidates. Across Europe, sovereign AI is moving from policy discussion to executable policy action.