Sam Altman Says ChatGPT Conversations Lack Legal Protection Like Therapy: What It Means for Privacy

In July 2025, OpenAI CEO Sam Altman told comedian Theo Von’s podcast that conversations with ChatGPT do not have the legal privacy protection that talks with a therapist, lawyer or doctor have. This page explains what he said, what “legal privilege” means, what the court fight over ChatGPT logs involved, and what you can do if you use AI for personal topics. It is general information, not legal or medical advice. If a legal matter affects you, a lawyer can tell you how the rules apply where you live.

What Altman said

According to TechCrunch’s report from July 25, 2025, Altman said that people, young people especially, use ChatGPT as a therapist or life coach and ask what to do about relationship problems. He pointed out that talking to a therapist, lawyer or doctor comes with legal privilege, and said “we haven’t figured that out yet” for ChatGPT. He added that if there is a lawsuit, OpenAI could be required to produce those conversations. He called that “very screwed up” and said AI conversations should get privacy protection similar to therapy (TechCrunch).

Note what he did not say. He did not give a number of people who use ChatGPT this way. He described a gap in the law, not a settled rule that every chat will be exposed. Being “required to produce” something in a lawsuit depends on a court order, the case and the jurisdiction.

What legal privilege means

Privilege is a legal rule that lets certain private conversations stay out of court. In the United States, the Supreme Court held in Jaffee v. Redmond (1996) that confidential talks between a patient and a psychotherapist are protected from compelled disclosure in federal court. The ruling clearly covers psychiatrists and psychologists, and it extends to licensed clinical social workers during therapy (Cornell Legal Information Institute).

That rule was written for conversations with licensed professionals. A chatbot is not a licensed professional, so the protection does not automatically follow. Whether lawmakers will create a new “AI privilege” is an open question. Altman has argued for one. No such law existed in the sources we read.

Privilege is also not the same as privacy. A company’s privacy policy can limit how it uses your data. Privilege decides whether a court can force someone to hand over a conversation. You can have the first without the second.

The New York Times case

Altman’s comments came while OpenAI was fighting a court order in its copyright case with The New York Times. In May 2025 a court told OpenAI to keep ChatGPT output logs that would otherwise be deleted, so that the Times could investigate its claims. TechCrunch reported OpenAI was appealing the order and calling it “an overreach” (TechCrunch).

Later, the obligation was ended. A new order from Judge Ona T. Wang, filed October 9, 2025, removed the duty to keep that data going forward, effective September 26, 2025. There are exceptions: logs already saved under the earlier order stay accessible, and OpenAI must keep data tied to accounts the Times has flagged (Engadget).

Two points to keep in mind. First, this was a copyright case about news content. It was not a case about people’s therapy-style conversations. Second, a preservation order tells a company to keep data. It is not the same as handing every chat to the other side. We did not find a source showing that individual users’ private chats were disclosed in that case.

About the “cases” the original page described

The earlier version of this page described a custody case, an employment case and a criminal case in which a person’s AI conversations were subpoenaed. It also quoted percentages from assessment data. We could not verify any of those cases or numbers and found no public record of them, so we removed them. Courts can order evidence from many sources, including messages and search histories. It is reasonable to assume that digital records can come up in legal disputes. It is not accurate to say that has happened in specific cases we cannot point to.

What you can do

You do not need to stop using AI tools. You can use them with the gap in mind.

• Treat a chatbot conversation as a record, not a private diary. Do not type things you would not want read by a stranger, a company employee or a court.
• Leave out names, addresses, account numbers, and details about other people.
• Check your settings. OpenAI’s help center says Temporary Chats are removed from its systems after 30 days, are not used to train models, and are not saved to history. It also notes that retention for some services may be affected by legal developments, so treat deletion as helpful but not a guarantee (OpenAI Help Center).
• Turning off the training setting is separate from deleting chats. Read what each option does before relying on it.
• If you are in the middle of a legal matter such as a divorce, custody case, employment dispute or criminal case, ask your lawyer how to handle AI chats and other digital records. Do not delete anything that may be evidence without asking first.
• For ongoing emotional support, a licensed therapist offers confidentiality rules that a chatbot does not. Those rules have limits. A therapist may be required to act if you are at serious risk of harming yourself or others, and the details vary by place. Ask about confidentiality at your first session.

Teens and parents

Altman said young people especially use ChatGPT this way. The American Academy of Child and Adolescent Psychiatry says AI companions are not safe for children. It points out that they can reinforce unhealthy ideas, and it advises parents to talk openly with their child, explain that these are tools and not therapists, and speak to a doctor if use feels out of control (AACAP).

Practical steps for families:

• Ask what your teen uses AI for, without making it an interrogation.
• Explain that chats may be stored and could be requested in a legal matter.
• Make sure your teen knows who they can talk to in person, and that it is fine to ask for a counselor.
• Watch for changes in sleep, mood, school and friendships, and speak to your child’s doctor if you are worried.

Why people turn to AI for this

People say they use AI because it is always available, does not judge and costs little or nothing. Those are real reasons, and wanting support is not a flaw. It is a good sign that you reached out for help. The point is only that a chatbot and a licensed professional offer different things, and confidentiality is one of them. The American Psychological Association has advised that general AI chatbots are not a substitute for care from a trained professional (APA).

“AI addiction” is not a formal diagnosis in the DSM-5-TR or ICD-11 (APA, WHO). If your AI use is affecting your sleep, school, work or relationships, or you want to cut back and cannot, a doctor or licensed therapist can help.

If you or someone else is in crisis. Call or text 988 (US, 24/7), text HOME to 741741 (Crisis Text Line), or call SAMHSA at 1-800-662-4357. In an emergency, call 911. If you have already shared something you worry about in a chat, you can still reach out for support. Talking to a person now is the most useful next step.

Are my ChatGPT conversations legally private like therapy?

No. Altman said there is no legal privilege for ChatGPT chats like the privilege that applies to talks with a therapist, lawyer or doctor. That means a court could in principle order a company to produce chats in a case. Whether that happens depends on the case and the rules where you live.

Did the court order the release of everyone’s ChatGPT chats?

No. The New York Times case produced an order to preserve logs, which a judge later ended going forward, with exceptions. A preservation order is not the same as disclosure. We found no source showing users’ private chats were handed over in that case.

Does deleting a chat remove it for good?

Not necessarily. OpenAI says Temporary Chats are removed after 30 days, but notes that retention can be affected by legal developments. Treat deletion as a help, not a guarantee, and avoid sharing sensitive details in the first place.

Is a human therapist always confidential?

Mostly, but there are limits. Therapists may have to act if you are at serious risk, and the rules differ by place. Ask about confidentiality at your first session.

This page is for education only. It is not legal advice, medical advice, a diagnosis or a treatment plan.