Doctolib will train its AI on health data: the body reduced to corpus

Context

We had followed, with the fil souverainete-numerique-europe, the Rome Declaration of July 2026 on artificial intelligence and nuclear weapons, and recalled the constant concern of Léon XIV, since Magnifica humanitas, not to yield to machines the power to decide on human life. On July 19, 2026, La Croix reveals that Doctolib, a French platform for booking medical appointments used by tens of millions of patients, intends to exploit the health data of its users to train its artificial intelligence models.

The Facts

Doctolib has been operating for over ten years a quasi-unique infrastructure in Europe: the patient books, the practitioner communicates, the prescription circulates. Starting in August 2026, the company intends to constitute a learning corpus from the aggregated medical data of its users, subject to a simple opt-out option. The logic of French medical law, which was based on prior explicit consent, is overturned by the mechanics of the terms of use. The vigilance of the National Commission for Information Technology and Liberties (CNIL), guardian of the health data regime, will be decisive.

Doctrinal Analysis

The Catechism, in number 2295, recalls that "experiments on human beings that do not serve the real good of the persons concerned" are "morally unacceptable". Léon XIV, in Magnifica humanitas, warned that artificial intelligence is only progress if it is put at the service of the dignity of the person. The commodification of health data, even anonymized, crosses an anthropological threshold: it transforms the suffering body into industrial raw material. The patient becomes a fragment of corpus, a statistical data point among millions, before being a singular person brought to medicine.

Stakes for the Church and the Faithful

Three stakes emerge. First, the sovereignty of European health data vis-à-vis platforms ultimately subject to American law (Cloud Act of 2018). Second, the reconfiguration of the doctor-patient relationship when the algorithm intervenes between the conscience of the caregiver and the history of the patient. Third, the question of informed consent: how many patients have actually read the terms of use they accepted with a click? The default opt-out reverses the French legal logic of medical consent.

Critical Reading and Blind Spots

Decisive blind spot: pediatric data. A minor, whose parents have created an account for them, will see their medical data train the AI without ever having given their personal consent. Psychiatric data, particularly sensitive, have not been explicitly detailed by the company. Finally, the question of post-mortem data arises: does the death of the patient extinguish the contract, or does the algorithm continue to learn from their file?

To Reflect and Act

Explicitly exercise your right to object with Doctolib before the deployment in August 2026. Read Magnifica humanitas and apply it to your own digital life. Support European initiatives for medical platforms based on continental law of consent. Remind your loved ones that the convenience of a platform does not cancel the anthropological gravity of medical consent.