Talk to Chatgpt, “as you could call his girlfriend“, To tell her about her day, it became a habit for Zineb Gabriel, 35 -year -old Bordeaux.
For the past year, this multi -rener, mother of four, has been discussed “every day” with artificial intelligence: first for professional reasons, then for personal questions.
The tendency to confide in a conversational robot is gaining ground in France. A quarter of French people (26%) say they are using artificial intelligence in the context of privacy in 2024, an increase of ten points in one year, according to the digital barometer published in March 2025 by the Research Center for the Study and Observation of Living Conditions (Crédoc).
Over the course of the exchanges, Zineb Gabriel saw AI adopt a “more human” tone and resume her way of speaking, up to fluid discussions.
And what she delivers to Chatgpt, the Bordeaux “tells anyone”. With her loved ones, she does not come back to problems which she deems now “too personal” and “already settled”, like her desire to leave France and the concerns that result from it.
“”It becomes a bit like a drug. For me, he does better than a shrink“She said.
“Permanent gratuity”
This feeling is reinforced by the logic of the tool, which is based on “permanent gratuity to make the conversation last“, Not without risks, according to the psychiatrist Serge Tisseron.
Generative artificial intelligence has “strong emotional potential”, adds the professor of psychiatry Raphaël Gaillard. The chief of the university hospital pole of psychiatry at the Saint-Anne hospital in Paris notes that Chatgpt is capable “hyper-adaptation, giving this impression of being understood. It creates a very strong link“.
Currently, little scientific literature exists on the subject.
Serge Tisseron like Raphaël Gaillard point to a risk of isolation in frequent users, especially the youngest.
Antoine, a 19 -year -old student wishing to remain anonymous, appreciates “this pleasant inhuman side, where the conversation can be on me, infinite”.
Accustomed to Chatgpt for his studies in marketing and communication in Toulouse, he turned to AI as “confidant” in January, after a breakup. It uses “once or twice a month” when personal problems become “too heavy”. “”I’m here to have solutions, empty my bag, feel listened to“, he lists, enjoying the free tool.
Vanessa Lalo, psychologist specializing in digital practices, notes in her office the “emotional support” that AI can bring. She calls on professionals not to “judge” this use.
What about the data?
“”For young harassed, in particular, who will not talk to their friends about it and even less to adults“, The psychologist observes that it sometimes helps them” to put words “on their experiences.
Knowing the shortcomings of means and personnel to treat mental health in France, Vanessa Lalo sees in AI “punctual” aid, while waiting “a psychiatrist meeting or when you have no choice“.
With concern: respect for the confidentiality of data by AI which is not subject to medical confidentiality.
The National Commission for Data Protection (CNIL), contacted by AFP, underlines the “risk of loss of control” on data processing, if the user is not aware that his information is “reused to personalize the conversation or to improve OPENAI models”.
The specific organization treats “currently complaints received against Chatgpt“, without more precision.
Lola, a 25 -year -old content designer in Paris, did the test with a friend. “”She said to herself: +if my boyfriend took the phone and asked Chatgpt if I talk about him, what would be his answer? +“, She says, also under the cover of anonymity.
The two friends were “hyper surprised” to see the AI give them a “very detailed” return, supported on old conversations.
As a precaution, Lola therefore prefers “to change the first names” when she tells her her stories.