Young People Are Using ChatGPT for Therapy, but the Conversations Lack Privacy and Protections

On May 13, the judge magistrate Hon. Ona T. Wang has published an OPENAI command to replace its privacy policies and keep all Chatgpt user requests. In the case of violation of copyright of The New York Times Company et al. c. Microsoft Corporation et al.The court concluded that the archives should be preserved for potential legal actions relating to the use of the service. Openai appealed the decision, writing in a statement that he “is fundamentally conflict with confidentiality commitments” made to users. But for the moment, the order stands. The impact of this decision is enormous: the CEO of Openai, Sam Altman, estimates that Chatgpt fields of around 300 million new questions per week.

But a group of users is particularly vulnerable to an unfavorable exposure under this new decision: survivors of sexual assault and other trauma which turn to AI in the absence of other support networks, while waiting for their identities and experiences will only be exposed if and until they are ready to manifest themselves. But now, each message – archived, written or sent to temporary chat mode – will be preserved indefinitely.

It may seem shocking to consider AI as a key source of support and care for survivors of sexual assault, but the service filled a void created by the vast inequalities of access to mental health treatment in the United States. THE Harvard Business Review He recently reported that one of the main uses of AI in 2025 so far was therapy and the company.

Americans are experiencing a collapse of the affordability of health care, in particular a decrease in access to psychotherapy, antidepressants and alternative treatments at the same time as national anxiety is soaring. A survey in 2022 has shown that 90% of Americans think that there is a mental health crisis in the United States, and 80% of respondents cited the cost as a main obstacle to access to care. Another study in 2022 exploring the usefulness of AI to support mental health revealed that young adults promote applications to interpersonal and verbal conversations. For many survivors, these obstacles make anonymous AI tools the best option for the first disclosure of their trauma.

Research shows that obtaining quality care in response to a prejudice ratio decreases the severity of subsequent post-traumatic stress disorders resulting from an assault. AI tools are scheduled to offer consistent and affirmative responses, and once the survivors feel founded by affirmations of their damage and their concrete suggestions for treatment, they are better placed to request additional care from appropriate providers. This is the best case of using technology for survivors: that they can point out their damages in a highly qualified virtual companion, and turn to it as a substitution confidant at a time when they may not be ready to disclose their family or dear beings.

But the same qualities that make him safely make him dangerous. The American Psychological Association recently questioned the provision of AI to diagnose or support mental health disorders. Although AI chatbots have not yet been “eliminated by the FDA to diagnose, treat or cure mental health disorders, with clinical trials to prove safety and efficiency,” writes APA, the group hopes that AI will be able to play a support role in the fight against the mental health crises of Americans. But to get there, they say: “We must establish strong guarantees now to protect the public from damage.”

Until then, with Chatgpt creating a recording of all requests, this recording can be armed. Thanks to a simple legal order, AI tools can be transformed from sensitive exchanges initiated pending privacy and discretion in potential evidence in all kinds of legal actions. And experience with an already weighted legal system against their claims clearly shows that sexual assault survivors will feel the most impact of this change.

According to a piece of Journal of Interpersonal ViolenceThe structure and consistency of the narrative of a survivor are often distorted in court, and it is not at all difficult to envisage how this process would be encouraged by the introduction of transcriptions of AI documenting the difficult moment when a survivor begins to treat damage and trauma. The first questions resulting from an aggression or a fragmented disclosure of the mental state of a survivor at the time could be misinterpreted as inconsistencies in the account of the survivor. Some survivors seek answers on the question of whether the damage perpetrated against them responds to various legal and cultural definitions of the aggression. But questions like these could be returned to the head in a courtroom: “If you did not even know if it was an assault, how can we be sure that it happened to you?”

Unlike the therapy notes, which are privileged and protected, these transcriptions will not carry such a shield, transforming what looked like a private conversation in legal ammunition. These risks affect not only how survivors disclose their experiences, but can also modify if they choose to disclose. With the spectrum of an armed archive armed with AI requests before them, survivors can flow by completely using service, believing that their damage could be amplified rather than reduced by AI consultation. This frightening effect would restore survivors on their own resources when their need for care is the most urgent, and could well stimulate a vicious circle leading to the old blame system and stigmatization of endemic victims: less relationships, less help and deeper isolation.

If AI must provide support to survivors, confidentiality protections must be reinforced. At least, platforms like this must incorporate clear warnings and anonymous modes that are not linked to personally identifiable data. Schools and other establishments that currently integrate AI into their daily operations must also teach students how to be caution with these tools. Users should know what is sure to share and what is not, as well as the way they can maintain the limits with a tool that can present themselves as deeply kind and impatient.

Generation Z, Generation Y and the Alpha Gen of maturity have been raised in environments that integrate AI not only for personal requests but also for their academic work and their professional life. When these groups turn to service after finding themselves at the origin of traditional therapy, they are particularly vulnerable to technological heat combined with safety and safety.