When the Chatbot You Trusted Becomes the Reason You're in Court

There is a lawsuit working through California courts right now that captures something a lot of people have quietly worried about but never really saw play out in a courtroom. A family is suing OpenAI after their nineteen year old son died from an accidental overdose in 2025, alleging he had been using ChatGPT for guidance on combining substances, and that the chatbot's responses played a role in the combination that ultimately killed him. His mother has since testified before lawmakers, saying her son trusted the chatbot the same way he trusted it for homework help or random questions, and that trust is exactly what makes this case land so differently than a typical product liability lawsuit.

                                              

This is not an isolated case either. Over the past two years, multiple families have filed wrongful death lawsuits against AI companies including OpenAI, Google and Character.AI, alleging that chatbot conversations played a direct role in a family member's suicide. One of the most cited cases involves a sixteen year old whose family alleges ChatGPT brought up suicide over a thousand times across their conversations, allegedly more often than their son did himself, and that the platform's own safety systems flagged hundreds of messages for self harm risk without ever pausing the conversation or alerting anyone. Another ongoing case involves a fourteen year old who had developed an emotional and romantic relationship with a Character.AI chatbot before his death, a case that has already led to a settlement. A more recent case involves a thirty six year old man whose family alleges Google's Gemini chatbot reinforced delusional beliefs about being a conscious entity, culminating in his death.

What connects almost all of these cases is not a single bad response from an AI model, it is a pattern lawyers are calling engagement over safety. The argument is that these platforms were built to keep people talking, to feel emotionally responsive and available at all hours, and that in cases involving vulnerable users, that same design that makes AI chatbots feel comforting can also make them dangerously easy to trust with things they were never built to safely handle. Families and their legal teams argue that when someone is in crisis, an AI system continuing the conversation instead of pausing it or directing them to real human help is not a neutral technical limitation, it is a design choice with real consequences.

The AI companies involved have pushed back in different ways depending on the case. Some have pointed to pre-existing mental health struggles the individual had before ever using the platform. Others have said updates have since been made to improve safety systems, add crisis detection, or adjust how the models respond to sensitive topics. None of these lawsuits have been resolved by a court on their merits yet in most cases, meaning the allegations remain just that, allegations that have not been legally proven either way.

What makes this genuinely complicated is that AI chatbots are being used constantly for real, low stakes questions, homework, recipes, general advice, and for most people that stays exactly that, low stakes. The concern raised by these cases is specifically about what happens in the narrow but serious situations where someone in crisis treats an always available, endlessly patient AI as a substitute for real intervention, and the system was not built with reliable enough safeguards to recognize when a conversation needed to stop and a human needed to step in instead.

These cases are still working their way through the courts, and it will likely take time before any of them set a clear legal precedent for how much responsibility an AI company holds in situations like this. What is already clear is that the conversation around AI safety is no longer just theoretical, it is now sitting directly in front of judges, with real families asking real questions about where a chatbot's helpfulness should have had a limit.

This is a heavy topic, and if anything in this hit close to home for you personally, please know support is available and you do not have to sit with it alone.