The French government has signed a new public contract worth 6 million euros with Mistral AI, a prominent artificial intelligence startup. The agreement, set to run from 2026 to 2027, involves the deployment of Mistral AI engineers within various government ministries to support three major strategic projects. These projects aim to enhance public services and strengthen the government’s digital capabilities. The initiative follows a significant data breach earlier this year that impacted the Direction générale des finances publiques, exposing the personal information of over 678,000 individuals. This incident has heightened the urgency of improving cybersecurity across public institutions. The three main areas of focus are cybersecurity, the detection of complex tax fraud, and the automation of dispute resolution processes within the Ministry of Justice. These efforts are intended to reduce the burden on the judicial system and improve efficiency. The technical implementation of these projects will be tailored to the sensitivity of the data involved, with options to use either Mistral AI’s own infrastructure, the state’s internal data centers, or the sovereign cloud platform Outscale. This flexibility aims to ensure both security and compliance with national data protection standards. Despite the government’s interest in Mistral AI’s capabilities, political leaders have expressed caution about becoming overly reliant on a single private company. During a recent visit to Silicon Valley, Economy Minister Roland Lescure highlighted concerns about depending too heavily on Mistral AI, which is valued at nearly 12 billion euros. He warned that if European AI development is seen as dependent on a single firm, it could hinder broader European innovation. “If European AI is summarized as Mistral, then we are doomed,” he said, emphasizing the need for a more diverse and robust AI ecosystem to compete with American and Chinese tech giants. Lescure also raised questions about Mistral AI’s future direction, noting whether the company will focus more on developing fundamental AI models or shift toward providing efficient, profitable infrastructure services. Additionally, he pointed to recent geopolitical tensions in the tech sector, such as the temporary blocking of access to models from the American company Anthropic for foreign users. This incident, he said, underscores the “unpredictability” of relying on foreign AI technologies, reinforcing the need for Europe to build its own resilient and independent digital infrastructure.