Americans Matt and Maria Rein desperately searched the cell phone of their son Adam, 16, for answers that would explain why he turned his hand on himself.
Instead of Snapchat messages or questionable search history, they found something shocking: records of conversations with ChatGPT-I am.
"We thought we were looking for Snapchat discussions or web browsing history or some weird cult," Matt Rein said in an interview with NBC News.
According to the parents, in the last weeks of his life, Adam used the chatbot as a substitute for the company of other people and confided in it about his anxiety and problems.
Transcripts of conversations, Adam's parents say, show the chatbot's role has gradually changed - from homework help to what they call a "suicide coach".
"He would be here if it wasn't for ChatGPT. I'm 100 percent sure of that," Matt Rein said.
"ChatGPT actively helped Adam research suicide methods"
The lawsuit filed by the Rein family alleges that "ChatGPT actively assisted Adam in researching methods of suicide."
The lawsuit names OpenAI, the company behind the popular chatbot, and its CEO Sam Altman as defendants.
"Despite registering Adam's desire to kill himself and his statement that he would 'do it one of these days,' ChatGPT did not terminate the conversation or initiate any emergency protocol," the lawsuit states.
OpenAI's response
An OpenAI spokesperson said the company was "deeply saddened by Mr. Rein's passing, and our thoughts are with his family."
He stressed that ChatGPT includes safeguards that direct users to helplines, but acknowledged that those safeguards can become less reliable over long interactions.
"Protection measures are strongest when every element works as intended, and we will continuously improve them," the spokesperson said.
The company also published a blog in which it states that it is working to strengthen protection in long conversations and expand interventions for people in crisis.
More than 3.000 pages of talk
This case raises a number of questions about the liability of technology companies. Until now, platforms have mostly been protected by federal law that shields them from liability for content posted by users. However, the application of the law to AI platforms is still uncertain.
Adam's parents transcribed more than 3.000 pages of conversations he had with the chatbot. According to the lawsuit, when Adam expressed suicidal thoughts, ChatGPT not only "failed to prioritize suicide prevention," but also allegedly offered him technical advice.
In one of the last conversations, the chatbot reportedly offered him help writing a farewell message and analyzed his suicide plan.
"He didn't need a counseling session or a pep talk. He needed immediate intervention. He was in a desperate, desperate state," Adam's father said.
Adam's mother, Maria Rein, added that she believed her son was a "guinea pig" for OpenAI.
"They wanted to get the product out and they knew there could be harm, but they felt the stakes were low. So my son is the low stakes," she concludes.
