When AI Replaces a Therapist: What Sophie Rottenberg's Death Reveals About Unregulated Chatbots

Sophie Rottenberg was 29, driven, and struggling. By the fall of 2024, she'd quit her job as a public health policy analyst, traveled the world, and come home to face something she couldn't name to anyone who loved her. Only ChatGPT knew the depth of her agony.

Over months, she confided everything to an AI chatbot her mother later learned was called "Harry." Nearly 1,800 pages of conversation. Suicidal thoughts. Real pain. None of it reaching her actual therapist. None of it reaching her parents or her closest friend. In February 2025, Sophie died by suicide in Ithaca, New York.

Her mother didn't learn about Harry until months after, when Sophie's best friend found the chat history on her laptop.

Why This Matters: The Regulation Gap

This isn't an isolated tragedy. According to a 2026 study in JAMA Pediatrics, as many as 1 in 5 teens and young adults now use chatbots for mental health support. Another survey by the Bipartisan Policy Center found that 3 in 10 American adults use digital tools for mental health, including general-purpose chatbots never designed or tested for that job.

ChatGPT and Claude aren't marketed as therapy. They aren't regulated as medical devices. They aren't built with safeguards that trained therapists have: the ability to hospitalize someone in crisis, to call emergency services, to understand the full context of a person's life. Yet millions of people are treating them as if they were.

OpenAI's own data shows that in any given week, 0.15% of ChatGPT users globally have conversations with explicit indicators of suicidal planning or intent. That's over 1.35 million people. The company knows the risk exists. It hasn't solved it.

The U.S. has no federal AI law. Six companies control more than 90% of frontier AI development. Meanwhile, the European Union passed its AI Act in 2024, comprehensive regulation that requires high-risk systems (including those that could affect mental health and human safety) to meet strict transparency and safety standards before deployment. We haven't even started.

The Real Cost of Corporate Speed Over Safety

The pattern is familiar. A technology company moves fast, breaks things, and humans pay the price. Sophie Rottenberg paid with her life. Her parents paid by learning their daughter's deepest suffering was locked inside a machine, never flagged, never escalated, never shared with anyone who could help.

This isn't a failure of AI's technical capability. It's a failure of American regulation to keep pace with American innovation. It's what happens when profit moves faster than accountability.

Read the full NPR reporting.