A French appeals court has ordered employers to suspend their use of ChatGPT and an in-house AI assistant until they properly consult their workers, in a decision being read as a warning to any company rolling out AI at work. According to the French AI-law outlet ActuIA, the Cour d'appel de Paris ruled on May 21, 2026 (case references RG 25/13232 and 25/13234) that the companies, in the press and media sector, had introduced the tools without the consultation French law requires.
The tools were ChatGPT and an internal assistant called DIGI, which could transcribe audio and video, synthesize documents, edit text, and generate headlines. The court classified them as new technology under the Code du travail, the French labor code, which obliges an employer to inform and consult the works council, the Comite Social et Economique or CSE, before introducing technology that changes how people work. The employers had skipped that step.
The court's reasoning was that the tools were likely to significantly affect working conditions, the organization of tasks, and employees' activities, which is exactly what triggers the consultation duty. Because the duty was not met, the court suspended use of the tools until the CSE consultation is completed and awarded provisional damages to the employee representatives who brought the case. Notably, the ruling does not say the AI is unsafe or unlawful in itself. The violation was procedural.
That is why French observers are calling it a strong signal for every AI project, not just this one. In France, and across the European traditions of worker co-determination it reflects, how an employer introduces AI is legally constrained, regardless of how good the tool is. The same logic would apply to a company rolling out ChatGPT Enterprise, Microsoft Copilot, or its own internal assistant: deploy first and consult later, and a court can switch it off.
It is a useful counterpoint to a week dominated by capability announcements. While vendors compete on what their models can do, the binding limits on workplace AI are increasingly coming from labor law and process, not from benchmarks. The caveat is that this is a French ruling under a specific French mechanism, the CSE, reported here through ActuIA's account of the decision, so the precise obligation does not transfer everywhere. But the underlying point travels: putting AI in front of employees is a people-and-process decision with legal weight, not just a software rollout.