France Travail's AI development

France Travail is creating a generative AI tool, similar to ChatGPT, to monitor over 6 million people registered with the organization. The tool is expected to be used in the first half of 2026 and will target categories A to E of Inscrits à France Travail. This has raised concerns about the use of AI for social control.

AI Ethics Committee announcement

On June 30, France Travail leaders announced in the AI Ethics Committee that they would develop a generative AI for employment search monitoring teams. The goals include "implementing continuous monitoring," "detecting weak signals," and "systematizing controls." This project highlights the normalization of AI in social control.

AI deployment in stages

The AI will be rolled out in stages. In the first phase, it will not trigger controls or interact directly with monitored individuals. Instead, it will support the profiling algorithm used to select people for monitoring. Controls can be targeted, random, or based on selection algorithms.

Short-term goal of AI

The short-term goal is to reduce the time needed for controls by shifting the analysis of dossiers to AI. The AI will generate a "structured summary" and "points of attention" to prepare "conclusions" of the control. Monitors will then validate the application of potential sanctions.

Long-term AI objectives

By 2030, AI aims to automate the entire control process, from selecting people to be monitored to executing the control itself. This involves giving AI access to data from all registered individuals and modifying its "prompt" to allow it to choose people to monitor and communicate with them via emails or SMS.

AI and control complexity

Generative AI represents a significant advancement in automating complex control operations. These involve analyzing multiple data points, considering various contexts, and communicating with monitored individuals. This automation could expand existing control methods and introduce new, more intrusive measures.

Tracking of journeys

The AI will be used for "tracking of journeys," which involves monitoring the obligation to complete 15 "hours of activity" per week. This requirement, introduced by the Full Employment Act, applies to all people registered with France Travail, regardless of whether they receive the RSA.

France Travail's response

France Travail argues that the AI will be a "decision-making aid," with the final choice of applying sanctions remaining with control teams. However, this is challenged by the potential for automation bias and the goal of increasing the pace of controls.

Ethics Committee meeting

On June 30, 2026, the France Travail Ethics Advisory Committee met again, with the agenda including AI use in employment search monitoring. This monitoring determines if unemployed individuals are actively seeking work and can lead to the suspension of unemployment benefits or removal from France Travail's records. Internal documents and testimonies collected by Mediapart reveal the scale of these projects, the flaws in the tools, and the lack of transparency in their development.