The French State Bets on AI to Modernize Public Services

The French State Bets on AI to Modernize Its Public Services

The government launched its new strategic plan “Our AI” on Tuesday, which could transform public services. The document discusses the deployment of sovereign tools to modernize administration and break dependence on tech giants.

The state’s “Our AI” plan, presented by the Minister of Action and Public Accounts on Tuesday, is quite comprehensive. Gone are the days of small artificial intelligence projects tested in secret within each ministry. On Tuesday, June 16, 2026, David Amiel announced the end of the every-man-for-himself approach by unveiling a large national plan to integrate AI throughout the administration. The dual objective is to relieve civil servants of repetitive tasks to free up time in the field, while also building 100% French technology to no longer depend on American or Asian software giants. This plan is accompanied by a generous budget of €655 million.

Artificial Intelligence Officially Enters State Ministries

The major asset of this transformation is simply called “The Assistant.” Tested for ten months by 10,000 public agents, this sovereign conversational robot is now accessible to all civil servants. It was designed by the State’s Digital Directorate (DINUM) in collaboration with the French company Mistral AI, and the tool is hosted on a SecNumCloud, the highest security level to protect citizens' data, guaranteed by ANSSI, the French cybersecurity agency. It can summarize complex files or draft letters in an instant, acting as a daily co-pilot to save precious time for teams.

The state’s “Visio” platform now integrates an AI capable of automatically recognizing who is speaking on screen to independently draft the meeting minutes. Meanwhile, the new tool “Transcripts” transforms any audio recording into dictated text with a single click. Thanks to this automation of voice recognition, administrative secretariats will save countless hours of tedious manual entry and free themselves from the heaviest transcription tasks.

Diplomacy and justice will also benefit from the deployment of “DiploIA,” an ultra-secure automatic translator capable of juggling 64 languages without any confidential data leaking abroad. For citizens, the revolution will come through the “Clear Language” application. This module uses text simplification algorithms to help tax agents (DGFiP) translate the administrative jargon of millions of letters sent to taxpayers into simple, understandable sentences.

AI, Yes, but the Administration Refuses to Replace Its Agents with Algorithms

However, the government emphasizes that it is not about replacing civil servants with cold, autonomous algorithms. The executive strongly insists on this point: AI must remain an assistant, and the state refuses to allow a machine to make administrative decisions alone, as seen in Nevada for processing local unemployment benefit applications. Moreover, a precise usage guide has just been published by France to strictly regulate the confidentiality of user data.

This somewhat abrupt transition is supported by excellent buy-in from the field, the government reassures. According to a wide array of internal surveys conducted among 2,000 civil servants, 83% of agents have a positive perception of AI. Even better, 59% believe it significantly reduces the burden of repetitive tasks and enhances their daily missions.

To support this movement without leaving anyone behind, a massive training plan will be rolled out by 2027. Starting in September, the Regional Institutes of Administration (IRA, currently being transformed into Public Service Institutes or ISP) will integrate AI into their curricula to prepare future state executives. Additionally, a major negotiation with unions will begin this autumn to establish an unprecedented framework agreement to protect working conditions and regulate the use of algorithms.

How the Government Plans to Reduce Its Financial Dependence on American Publishers Like Microsoft

To steer this digital ship, DINUM (the State’s Digital Directorate) is undergoing a profound reorganization under the “ARIANE” plan. Having become the trusted technological architect of the administration, it will centralize and share critical infrastructures, including the supercomputers essential for running AI. Its goal will also be to impose common purchasing rules across all ministries to prioritize French and European solutions, thereby directly supporting our industrial sovereignty.

The state has done the math! It has realized that it spends a staggering €4.5 billion annually on its IT. A significant portion of this public money flows to foreign non-sovereign tech giants, such as the €100 million paid each year to Microsoft for software licenses. To stop this capital and data leak, the plan imposes a strict roadmap on each ministry starting this autumn to gradually replace these software solutions with local alternatives. Once and for all?

The financial automation of the state is progressing thanks to the “Informed Spending” project, integrated into the public budgeting software Chorus, which will be rolled out by the end of the year. By automatically analyzing public procurement contracts, this AI will relieve agents of endless manual entry, allowing for productivity gains estimated by the ministry at 2,000 man-days per year. To sustain these gains and regain control of its technology, the state commits to re-internalizing tech jobs by 2027: every euro invested in hiring an IT professional or AI expert within the administration will permanently eliminate €1.50 in spending on external consultants and service providers.

By shaking up its own infrastructures in this way, the state imposes a true credibility test on itself. It remains to be seen whether this digital rearmament will be sufficient to make France the European model of a modern, efficient administration that is truly in control of its algorithms.