ChatGPT users may want to reconsider relying on the AI app for therapy or emotional support. OpenAI CEO Sam Altman recently highlighted that the AI industry has not yet established safeguards for user privacy in sensitive conversations. Unlike interactions with human therapists, doctors, or lawyers, there is no legal confidentiality protecting chats with AI.
Altman shared these concerns during an episode of Theo Von’s podcast, *This Past Weekend w/ Theo Von*. He explained that without a legal or policy framework for AI, user conversations lack the same privacy protections as those with licensed professionals. Many people, especially younger users, turn to ChatGPT for personal advice—discussing relationships, mental health, and life decisions. Currently, OpenAI could be compelled to disclose these conversations in legal cases, posing a significant privacy risk.
Altman called the situation problematic, arguing that AI interactions should have the same confidentiality as therapy sessions. He acknowledged that this issue was not a consideration just a year ago but is now a critical barrier to wider AI adoption.
OpenAI is already facing legal challenges regarding user data. The company is contesting a court order in its lawsuit with *The New York Times*, which would force OpenAI to retain the chat histories of millions of users globally, excluding enterprise clients. OpenAI has labeled the order an overreach, warning that compliance could set a precedent for broader data demands in future legal or law enforcement cases.
Privacy concerns extend beyond AI. After the Supreme Court overturned *Roe v. Wade*, many users abandoned period-tracking apps for more secure alternatives, fearing their data could be used against them. Similarly, Altman emphasized the need for clear privacy and legal standards before users fully trust AI with personal matters.
When asked about his own ChatGPT usage, podcast host Theo Von admitted he avoids the chatbot due to privacy concerns. Altman agreed, stating that clarity on privacy protections is essential before relying heavily on AI for sensitive discussions.