France Travail is testing an AI system intended to detect job seekers to be checked as a priority. To achieve this, the algorithm analyzes 26 variables.
The media La Quadrature du Net, in partnership with the investigation unit of Radio France, revealed an internal document from France Travail detailing the experimentation of an algorithmic system intended to profile job seekers. The objective: to achieve the 1.5 million checks announced by Gabriel Attal by 2027. If France Travail promises an ethical model, the risks of AI abuse and the inevitable lack of human supervision are cause for concern.
France Travail will profile job seekers using AI
The France Travail project has the following name: “Targeting job search control”. To be more precise, and above all clearer, it is a question here of a profiling using artificial intelligence of people looking for a job. The public employment service explains that the objective of the project is to “ use AI » in order to « identify files requiring in-depth review » and “ automatically populate the job search monitoring tool ».
To analyze each job seeker file, the AI is based on “ 26 key variables “. We can cite in particular the declared activities, recent interviews, the visibility of the job seeker with recruiters, the level of training, the intensity and types of services mobilized, the history of breaches or even the age of registration. Based on these key variables, the algorithmic model should detect people who would fail to fulfill their obligations job search, placing them either in a “suspect” category or in the “non-suspect” category. Those who are placed in the first category will therefore experience priority control. Let us point out that, for the moment, this artificial intelligence does not issue any sanctions.
Let us also specify that France Travail wishes “ extend the use of the model to other audiences » in order to « automatically populate the control tool and transmit targeted checklists each month “. As La Quadrature du Net points out, if we take into account all the RSA beneficiaries, now forced to register with France Travail since January 1, 2026, there are more than 6 million people which could be evaluated each month by the algorithm.
Mass surveillance and a risk of AI drift?
In 2017, the number of checks on job seekers reached 200,000 in France. This number rose to 944,000 in 2025. Gabriel Attal announced a target of 1.5 million checks for 2027. The French government therefore wants to strengthen the control of job seekers. Carrying out 1.5 million checks is difficult for human beings to do. The use of AI therefore seems inevitable. Likewise, given the number of files processed by the algorithmic system, human supervision risks becoming very complicated to say the least.
And even if France Travail ensures a “ removal of preconceptions » by proposing “ files to be checked on the basis of objective criteria », how to be sure that the AI will not present any deviant behavior ? How will AI be able to fairly and ethically judge a person on the basis of a few criteria, such as their visibility among recruiters or their level of training? So many questions that remain unanswered. La Quadrature du Net rightly criticizes France Travail for its lack of transparency concerning the operation of its algorithmic tool. Regardless, the process is now underway.
👉🏻 Follow tech news in real time: add 01net to your sources on Google, and subscribe to our WhatsApp channel.
Source :
France Travail déploie un outil de profilage algorithmique à des fins de contrôle
