OpenAI CEO Sam Altman has raised alarms over the lack of privacy protections in AI interactions, warning that conversations with ChatGPT could be subject to legal disclosure. His remarks have sparked renewed debate over data privacy in the age of generative AI.

In a recent appearance on the This Past Weekend podcast hosted by Theo Von, Altman pointed out that many people—especially younger users—are confiding in AI tools as they would in therapists or life coaches, often revealing deeply personal information.

“People are using it like a therapist, a life coach,” Altman said. “But if you talk to a doctor, lawyer, or therapist, that’s protected by legal privilege. If you talk to ChatGPT and there’s a lawsuit, we could be compelled to hand over that data.”

Calling the situation “very screwed up,” Altman emphasized the need for privacy protections for AI conversations, arguing they should be treated with the same confidentiality as interactions with licensed professionals. “We should have the same concept of privacy,” he added.

His comments come amid growing scrutiny from privacy advocates, who are increasingly concerned about how AI companies handle user data. While platforms like ChatGPT include safety mechanisms, a clear legal framework for user data does not yet exist—leaving interactions vulnerable to legal demands.

As of mid-2025, ChatGPT boasts between 800 million and one billion weekly active users—a dramatic increase from 400 million earlier in the year and 300 million in late 2024. The platform has also attracted roughly 10 million paying users and one million commercial subscribers. OpenAI is on track to surpass one billion users by year’s end.

With the platform becoming an integral part of people’s lives—emotionally, personally, and professionally—the stakes for establishing robust privacy protections have never been higher. Even Von, the podcast host, admitted he’s hesitated to use AI due to uncertainties about data security.

“I think that makes sense,” Altman replied. “It’s important to get legal and privacy clarity before relying heavily on it.”