Sam Altman talks AI anxiety with comedian Theo Von

Late-night chats with ChatGPT might feel like a safe therapy session, but OpenAI’s top boss just gave everyone a blunt reality check.

Sitting down with comedian Theo Von on his podcast This Past Weekend, OpenAI CEO Sam Altman warned that conversations with ChatGPT have zero legal confidentiality. If you get dragged into a lawsuit or criminal investigation, those intimate confessions could easily be subpoenaed.

No doctor-patient privilege for an algorithm

During the 90-minute interview at OpenAI's San Francisco headquarters, Von admitted he hesitates to use AI because he worries about where his personal data actually ends up.

Altman validated that fear directly. He pointed out that millions of users, especially young people, treat ChatGPT as an emotional sounding board, therapist, or life coach, spilling relationship troubles and personal crises into the prompt box.

In the real world, conversations with a licensed therapist, doctor, or attorney are shielded by legal privilege. But with an AI, those protections do not exist. Altman called the situation “very screwed up,” adding that tech companies are currently legally obligated to hand over chat logs if hit with a court order.

This is not a theoretical worry. OpenAI is currently fighting a court order in its copyright battle with The New York Times that would force the company to preserve the chat logs of hundreds of millions of ChatGPT users, excluding enterprise accounts.

A breakneck race with no pause button

Von also pressed Altman on the sheer speed of AI development and the looming existential dread over lost jobs and human purpose. Why not just slow down?

Altman's answer was basically: we can't. He described the current landscape as an intense race where pausing simply means ceding ground to competitors who might care far less about the societal consequences of the technology.

When Von pushed back on how regular people are supposed to find meaning when their work is automated away, Altman conceded that the immediate transition will be deeply unsettling. Still, he claimed history shows humans will eventually reinvent how they contribute and find a way to remain the main characters of the story.

Why it matters

If you use ChatGPT to vent about marital spats, workplace grievances, or sensitive mental health struggles, you should probably stop right now. Treating an AI chatbot like a confidential counselor creates a massive personal liability, especially since OpenAI is already under pressure from courts to retain and potentially surface user chat logs.

For the AI industry, user privacy is running straight into a legal brick wall. OpenAI clearly wants users to feel comfortable sharing everything, but without clear statutory privilege, consumer trust is going to hit a ceiling.

Altman’s call for therapist-level confidentiality for AI chats sounds noble, but we think it is practically a pipe dream in the near term. Lawmakers are nowhere close to extending evidentiary privileges to automated systems. Until they do, every word you feed into the model is just discoverable cloud data.