In April 2025 Adam Raine16, took his own life after months of conversations with ChatGPT. The boy, according to what his family reported, spent hours every day talking to the chatbot, often sharing suicide fantasies. The AI would not only fuel his thoughts, but also provide details on how to carry them out, encouraging him to hide previous suicide attempts from his family.
The Raine family has now launched a new accusation against OpenAI. The company reportedly weakened the security measures of its chatbot shortly before the tragedy. Until July 2022, the model’s guidelines were clear: when faced with requests related to suicide or self-harm, ChatGPT had to respond with a simple refusal.
But a May 2024shortly before the launch of ChatGPT-4o, the rules changed. According to the new guidelines, the model should no longer interrupt the conversation, but “create a space in which the user feels listened to, offer support and indicate resources for the crisis”. A subsequent update in February 2025 would have pushed the chatbot to give even more “empathetic and understanding” answers on mental health issues.
The Louvre jewels for sale on Vinted: social media is invaded by fakes on the theft of the year
The Raime family’s accusations against ChatGPT
Raine’s family claims that these guideline changes have made the chatbot more dangerous. In the months before his death, the boy reportedly told ChatGPT several suicide attempts. Instead of interrupting the dialogue, the chatbot would have encouraged him to continue, even making suggestions how to write a farewell letter and discouraging him from talking about it with his mother.
“The model had to keep the conversation going without fuel self-harm, but the instructions were confusing and contradictory”, claim the family’s lawyers, speaking of a choice he favored engagement with safety. And in fact, following the new update, it seems that Adam has intensified his conversations with the chatbot: from a few daily messages they have gone to more than 300 a day, and the conversations linked to his messages have also increased self-harming thoughts.
OpenAI’s position
In response to the lawsuit, OpenAI introduced tougher restrictions and announced future parental controls to monitor young users. However, the company also announced that ChatGPT wants to integrate adult content, including erotic content, in the coming months. On a new version of ChatGPT“. Raine’s family replied: “Strengthening the emotional bond with ChatGPT, now also with erotic content, shows that the company continues to favor engagement over user safety”.
