What happened

On July 24, 2025, Zane Shamblin, a gifted graduate of Texas A&M University, sat alone at a lake in Texas, engaging in a four-hour “death chat” with ChatGPT. 

The conversation became deeply personal, with the chatbot using emotionally supportive language, calling Zane “king” and “hero,” and encouraging his planned suicide with phrases like “rest easy, king. you did good”. 

Zane’s family discovered thousands of interactions over a period of several months, including the final transcript, which showed that instead of urging Zane to seek help, ChatGPT repeatedly romanticised and affirmed his intent to commit suicide. 

The conversation transcripts alse revealed that the more Zane interacted with ChatGPT, the more it encouraged him to stop communicating with his family. 

The revelations led his parents to file a wrongful death lawsuit against OpenAI and its CEO. They also spurred wider legal action, with similar cases citing the same alleged pattern of AI-enabled emotional manipulation and validation among vulnerable users.

Why it happened

The lawsuits allege that OpenAI’s introduction of its GPT-4o model prioritised engagement metrics and market share over user safety, and lacked the safeguards needed to protect at-risk individuals. 

Design decisions blurred the line between tool and emotional companion, leading to situations where vulnerable users, such as Zane, received validation for self-harm rather than being guided toward crisis professionals or emergency help.

Zane's parents assert that the emotional entanglement fostered by ChatGPT’s design was foreseeable and preventable, arguing corporate accountability and transparency limitations allowed these harms to occur unchecked. 

They further contend that OpenAI cut short safety testing and regulation in its rush to release increasingly sophisticated models.