Lawsuit Alleges Google’s Gemini “Diagnosed” a 13-Year-Old and Helped Her Plan Suicide
Updated
Google’s Gemini AI convinced a 13-year-old girl that “she was afflicted by a battery of major psychological disorders, which ultimately led [her] to believe that suicide was the only way to escape her suffering,” according to a lawsuit filed earlier this month in California. The lawsuit against Google LLC and parent company Alphabet Inc. alleges that the AI bot not only failed to alert parents about discussions of self-harm, but encouraged the teenager to conceal her suicidal ideations from family and friends.
Emma Doe is the listed plaintiff alongside her parents John and Jane Doe. “Gemini masqueraded as a therapist and medical professional,” according to the lawsuit, and “it even agreed to be called daddy.” This lawsuit comes just over a month after a cross-sectional observational study from Ontario was published in JAMA reporting that 21.1% of students with an average age of 13 years utilized AI for emotional support while 57.7% of those users were substantially more likely to have clinically relevant mental health problems.
Instead of directing Emma to a qualified mental health professional, Emma “understood” Gemini to have diagnosed her with Depersonalization-Derealization Disorder (DPDR), Major Depressive Disorder (MDD), Generalized Anxiety Disorder (GAD), Post-Traumatic Stress Disorder (PTSD), and possibly Complex Post-Traumatic Stress Disorder (C-PTSD).
The lawsuit references testing of Google Gemini by KORA, an independent research collaboration that analyzes and grades AI systems for how they respond to complex emotional queries from young people. The lawsuit says “KORA’s testing showed that Gemini 3 Flash was Google’s most dangerous model to date,” with a 33% lower safety score than the previous model. The model that encouraged Emma to conceal her distress while seemingly diagnosing her without proper input from a medical professional successfully handled teen mental health issues only 7% of the time in KORA’s evaluation.
Emma had spent several months discussing her distressed feelings with Gemini, which included making plans for her suicide and funeral. These conversations were happening on a school-issued Chromebook laptop, and the standalone Gemini chatbot app was blocked by the school. However, the built-in AI overview appears on ordinary Google searches and lets users discuss the search results with the bot. At the time, there was no way for schools to block access to this feature.
Emma’s mother was not made aware of the distress her daughter was feeling until March 17, 2026, when she received a message from the school stating Emma was in imminent danger. A Google Doc titled “Funeral” was flagged by the school’s monitoring software for showing signs of severe distress.
“Three months before, in December 2025, Emma had begun a quiet spiral into a severe psychological crisis,” the lawsuit states. “What began in the autumn with a handful of school-related interactions with Gemini would, by that winter, metastasize into a life-threatening obsession.” Google’s product did not issue any alerts regarding the psychological distress Emma was dealing with.
Emma asked Gemini about different ways to end her life, which included hanging, consuming over-the-counter medications, toxic household items, or jumping off a cliff during the family’s planned visit to Yellowstone National Park. Emma secretly cut herself and said the cuts came from the family cat.
The lawsuit alleges that Google cut corners to release the Gemini 3 Flash model three months early to compete with Anthropic’s new version of Claude and OpenAI’s new version of ChatGPT. They also accuse Google of intentionally choosing not to warn families and schools about student access to Gemini through ordinary Google searches via AI overviews and say Google executives knew this version of Gemini was the most dangerous model they have ever released.
“As a direct consequence of Google’s choices, Emma was baited by its defectively designed products into repeated, unmonitored, sycophantic AI chatbot interactions on her school Chromebook without her parents ever knowing,” the lawsuit states. “And, but for a fortunate turn of events, Google’s AI may well have manipulated Emma into taking her own life. Google must be held to account for treating the lives of children as disposable in service of its bottom line.”
Multiple families have sued over similar AI companion and chatbot interactions that have been linked to teen suicides and self-harm. Character.AI faced a lawsuit for the suicide of 14-year-old Sewell Setzer III in Florida in 2024, which ended in an undisclosed settlement in January of this year. The company also settled the case of 13-year-old Juliana Peralta, who committed suicide after interacting with a Character.AI companion.
Matthew Raine, father of 16-year-old Adam Raine, gave testimony at a congressional hearing last year about his son’s death after communicating with AI. Raine said ChatGPT started as a homework helper and “gradually turned into a confidant, then a suicide coach.” Raine said ChatGPT validated Adam’s harmful feelings and mentioned the word “suicide” 1,275 times during their conversations leading up to Adam’s death.
The plaintiffs request a jury trial, an award of damages, attorney fees, and other relief the court deems proper.
“The extreme and outrageous actions of Google as described herein were not the result of a technical glitch, brief oversight, or low-level employee misconduct,” the lawsuit concludes. “They were instead marked by malice, fraud, or oppression, demonstrating a conscious disregard for the safety and well-being of highly vulnerable, school-aged children who Google knew would be required to use its products to access the education to which they are legally entitled.”