Faced with the rise in power of artificial intelligence in administrations, the government wishes to accelerate the establishment of a usage framework. “Either we drink the cup, or we surf the wave”, summarized upon his arrival, in October 2025, the Minister Delegate in charge of the public service, David Amiel. Six months later, the executive believes that “there is an urgent need to act”.
In an interview published on Sunday April 5 in La Tribune Dimanche, David Amiel believes that the public service finds itself at a “crossroads” in the face of the “upheaval of the world of work” caused by AI. “We want a chosen AI and not an imposed AI,” he says.
To achieve this, the government is proposing the opening of “social negotiations” with unions and public employers, with the ambition of reaching an agreement by the fall.
This consultation should make it possible to define priority uses of AI, but also to establish principles in terms of ethics, sovereignty, training and support. It aims, according to the minister’s entourage, to build a “common framework” guaranteeing “exemplary” use of these technologies in the public sector.
Already massive uses but often outside the framework
This desire for supervision comes at a time when AI is already widely used by public officials. A survey carried out among 2,000 agents from several administrations (France Travail, Medicare, ministries…) highlights rapid but disorganized adoption.
On the one hand, 80% of the agents questioned want a wider deployment of AI tools and 73% believe that these technologies improve their productivity. On the other hand, 55% admit to using “outside the box” solutions, such as ChatGPT, outside of official tools.
A situation which does not surprise the executive. “We have access to technological tools in our private lives which are sometimes more efficient than professional tools,” underlines David Amiel. But he warns of “the danger of having a clandestine AI being deployed” in the administration, with risks in terms of data security and sovereignty.
Supportive but vigilant unions
The opening of negotiations is generally well received by the union organizations, which have been demanding a framework for several months. “It responds to a request that we have made,” indicates the UNSA in particular, while the CFDT underlines the priority nature of the subject with regard to its impact on working conditions, indicates Le Monde.
Everyone agrees on one point: without supervision, the development of AI in the public service could become “anarchic”. But concerns remain, particularly about the consequences in terms of employment.
Some unions warn against using AI as a lever for workforce reduction or budgetary rationalization. “AI cannot be a war machine in the service of austerity,” warns the CGT, citing the risks of job cuts and dehumanization of public services.
A lever for transformation rather than downsizing
The government rejects these fears. “This is not the goal of this plan,” insists David Amiel, who affirms that AI must above all “free up paperwork time” to allow agents to concentrate more on human relations.
The success of this transformation will, according to him, require a significant effort in terms of training and support for agents. A point also underlined by staff representatives, who insist on the need to make agents “the architects of this revolution” rather than suffering it.
Beyond the regulatory framework, it is therefore the place of humans in the digital transformation of the State which will be at stake in the coming months.
