San Francisco, July 27, 2025 –
In the age of artificial intelligence, millions of users worldwide are turning to tools like ChatGPT not just for assistance with writing, learning, or coding—but also for advice on deeply personal matters. However, OpenAI CEO Sam Altman has issued a word of caution to users who treat AI like a confidant or therapist: “Be careful what you share.”

Speaking on a recent podcast, Altman addressed growing concerns around data privacy, emotional reliance, and the limits of legal protection when users divulge personal information to AI platforms like ChatGPT.

The Emotional Bond with AI

As AI chatbots have evolved to become more fluent, empathetic, and responsive, a significant number of users—especially the younger generation—have started using ChatGPT for emotional support, therapy-like conversations, and personal dilemmas. From relationship advice to mental health questions, many users now treat the AI assistant as a digital friend or life coach.

But Altman believes that this behavior, while understandable, comes with potential risks.

“People are using AI tools like ChatGPT as therapists or life coaches, asking about relationship issues and emotional challenges. If you talk to a real doctor or lawyer about these things, you’re protected under specific legal rights. That’s not the case with AI,” Altman said.

Privacy is Not Absolute with AI

When asked by the host if he feels comfortable sharing personal information with AI tools, Altman replied, “I think it’s understandable to hesitate.”

He explained that while OpenAI is committed to safeguarding user data and has strict privacy policies in place, there are no legal protections comparable to attorney-client or doctor-patient privilege when interacting with AI models.

In other words, if a user shares deeply sensitive or incriminating information with ChatGPT, and if that data were ever requested through legal proceedings or investigations, there’s no legal barrier preventing it from being revealed.

“If you discuss the most sensitive topic imaginable with ChatGPT, and someday it becomes part of a lawsuit or has to be revealed, that could be devastating,” he added. “So, you should treat AI conversations with the same caution you would use in any potentially monitored or public setting.”

A Caution About Over-Reliance on AI

Altman also expressed discomfort with how deeply integrated AI has become in people’s daily lives. He said the growing dependence on AI tools for even minor decisions makes him concerned about the future of human independence and decision-making.

“People have started relying on ChatGPT too much, and it makes me feel terrible,” Altman said candidly. “If we begin living our lives purely based on what AI says, that’s both harmful and dangerous.”

He emphasized that while AI can be a powerful assistant, it should not replace human judgment, intuition, or emotional intelligence. Over-reliance on AI, according to Altman, can dull critical thinking and promote a kind of digital dependency that may be difficult to reverse.

What’s the Real Risk?

While companies like OpenAI continuously stress their commitment to data security, the risk isn’t necessarily about hacks or leaks. Rather, it lies in the legal gray area where AI companies aren’t bound by traditional confidentiality agreements. If a conversation were somehow involved in a legal case, and if authorities requested access, the data could potentially be disclosed.

This is especially concerning for those who have used ChatGPT to discuss:

• Legal matters or confessions
• Mental health struggles
• Relationship abuse or trauma
• Personal financial information
• Workplace harassment

While ChatGPT doesn’t store conversations by default in a way tied to individual users unless they opt-in to save chat history, users should still assume that no AI conversation is completely private.

What Can Users Do?

Given Altman’s own warning, here are some key takeaways for everyday users of AI tools like ChatGPT:

1. Avoid Sharing Deep Secrets: Don’t share personal trauma, legal confessions, or confidential workplace details with AI tools.
2. Use AI as a Supplement, Not a Substitute: AI can assist, but should not replace qualified professionals like doctors, lawyers, or therapists.
3. Understand the Limits of Privacy: AI companies have privacy safeguards, but no legal obligation to maintain strict confidentiality like a human therapist or legal advisor.
4. Practice Digital Hygiene: Turn off chat history if you’re discussing sensitive topics, avoid using real names or identifying information, and don’t assume your conversation will vanish forever.

AI Companies and Ethics

Altman’s candid comments come at a time when the ethical implications of AI are being hotly debated across industries. Questions about how user data is stored, used, or shared have prompted regulators worldwide to explore new data protection frameworks for AI interactions.

OpenAI has taken steps to ensure transparency and control, offering users the ability to delete chat history, opt-out of training models with their data, and toggle privacy settings. However, technology cannot yet replicate human ethical obligations, which is why Altman urges caution.