In a candid warning from the man behind ChatGPT, Sam Altman said that the personal data users share with the chatbot does not enjoy the same legal protections as conversations with a doctor or lawyer, and could be disclosed in legal proceedings.
Article contents:
- Introduction: the illusion of privacy
- Altman’s stark warning
- No legal privilege: the difference between AI and professionals
- A call for a new policy framework
- Mental health concerns
- Conclusion: think twice before sharing your secrets
Introduction: the illusion of privacy
If you are confessing your deepest secrets to an AI-powered chatbot, it might be time to reassess. As more people turn to artificial intelligence for instant life coaching, tools like ChatGPT absorb vast amounts of personal information about their users. While this data remains private under ideal circumstances, it can be subpoenaed in court—a scenario warned against by OpenAI CEO Sam Altman.
Altman’s stark warning
During an appearance on a popular podcast this week, Altman said: “One of the examples that we’ve been thinking about a lot… people talk about the most personal things in their lives to ChatGPT.” He added: “Young people, in particular, use it as a therapist, as a life coach. ‘I’m having these relationship problems, what should I do?'”
But he clarified the crucial difference: “Right now, if you talk to a therapist or a lawyer or a doctor about those problems, there is a legal privilege for it. There is doctor-patient confidentiality. There is legal confidentiality.”
No legal privilege: the difference between AI and professionals
Altman says that as a society, “we haven’t figured that out yet” for ChatGPT. “If you go talk to ChatGPT about your most sensitive stuff and then there’s a lawsuit or whatever else, we might be required to produce that. And I think that’s really bad.”
While interactions with doctors and therapists are protected under federal privacy laws in the United States, exceptions exist in cases where someone poses a threat to themselves or others. Even with these strong privacy protections, relevant medical information can be disclosed by court order, subpoena, or warrant. For AI conversations, such protection does not exist at all.
A call for a new policy framework
Altman argued that AI conversations should be treated with the same level of privacy as talking to a therapist. He said: “I think we should have the same concept of privacy for your conversations with AI as you do with a therapist.” This call for a new policy framework comes at a time when OpenAI and its peers have lobbied for a light-touch regulatory approach.
Altman also expressed concerns about how AI could negatively impact mental health, even as people seek out its advice instead of the real thing. He said: “Another thing that I fear… is just what this is going to mean for the mental health of users. There are a lot of people who talk to ChatGPT all day long.” He added: “I don’t think we know yet the ways that [AI] will have those negative effects, but I certainly feel like it’s going to have some. And we’re going to have to, hopefully, learn how to mitigate them quickly.”
Conclusion: think twice before sharing your secrets
Sam Altman’s warning serves as a powerful reminder that privacy in the digital age is a complex concept. While a conversation with ChatGPT may feel private and personal, technically and legally it is not. Users should be aware that their digital secrets could become public record, and think carefully about the information they choose to share with their AI companions.
Frequently asked questions
Does OpenAI read my chats?
According to OpenAI’s privacy policy, conversations may be reviewed by human reviewers to improve their models. You can opt out of this through your settings, but the company still retains the data and could legally be required to hand it over.
What is “legal confidentiality”?
It is a legal principle that protects communications between a client and their attorney from being disclosed without the client’s permission. Similar protection, called “privilege,” exists for doctor-patient and therapist-patient relationships.
What are the more private alternatives?
Talking to a licensed human professional (doctor, therapist) provides the strongest privacy protection. For AI, models that run locally on your device without sharing data with company servers may emerge in the future, providing better privacy.
