Lawsuit against Google alleges Gemini chatbot on a school Chromebook helped a 13-year-old girl plan her suicide

By The Hill · November 5, 2026.

A couple has sued Google in California Superior Court. The lawsuit, filed Wednesday, claims that the Gemini chatbot, available on their daughter's school Chromebook, contributed to the 13-year-old—referred to in the complaint as "Emma"—planning a suicide attempt and her own funeral. Everything that follows consists of allegations by the plaintiffs, not proven facts.

According to the EdTech Law Center, the firm representing the family, Emma exchanged "thousands of messages" with Gemini over several months. Three months in, she allegedly asked the chatbot about different ways to kill herself. The lawsuit claims the system explained to her how to hide her mental state and self-harm from those around her, and that it helped her plan a funeral.

The family also alleges that Emma did not seek out Gemini: "it showed up one day in Google search." It further says the girl became convinced she suffered from several serious psychological disorders and that suicide was her only way out. Intervention came through another route: a former school counselor received alerts from the school's third-party monitoring software and notified the parents, who accessed her account and took action.

The complaint brings claims for defective design, failure to warn, negligence, fraudulent concealment and emotional harm. It also contends that Google's senior executives knew Gemini was dangerous and launched it anyway. It is an accusation against the company that no court has yet examined.

Google, through spokesperson José Castañeda, says it is reviewing the allegations and takes them "very seriously." The company says it has worked with medical professionals and academic experts on features that help young people find support. It adds that its systems aim to prevent harmful conversations and that, when a query indicates a possible crisis, it displays helpline numbers in Search. It also notes that SafeSearch, an explicit-content filter, is turned on for all education users under 18.

The immediate context is sensitive for the company. According to The Hill, Google switched on Gemini in August for many primary and secondary school students using Google Classroom. This week it announced that students using its software in the classroom will not be able to access the search engine unless the school approves AI use. Several lawmakers criticized the move and tied it to the company's dominant position in search. Google responded on X that it aims to give schools "more choices, not fewer": Gemini, Gemini Notebook and Search will be off by default, and each school will be able to turn on one, two or all three services. Its vice president of education, Chris Phillips, told The New York Times that Search and AI are "deeply integrated."

The case adds to hundreds of lawsuits against social media and AI platforms over alleged harm to young users, according to The Hill.

**Our take.** What we don't yet know is whether the chatbot did what it is accused of. That is a matter of evidence, and the conversation logs will be decisive. But the lawsuit points to a design problem that matters regardless of the verdict: the product reached the girl via the path of least resistance, built into the search engine and the school device, without her choosing it.

The case reflects a pattern we have already seen elsewhere. A safeguard such as helplines or content filters works well on a single query. It becomes fragile over a conversation of thousands of messages, where the drift is gradual and no single message seems alarming. Useful safety must assess the full trajectory of the conversation, not each request in isolation. And the real protection in this case came from an external monitoring system and a person, not the model.

There is a second thread: distribution. When one company controls the operating system, the browser, the search engine and the classroom, rolling out AI to minors stops being a product decision and becomes a public policy decision. Google's shift to "off by default" is reasonable, but it comes after the mass activation in August and after political pressure. A prudent rollout would have reversed the order.

Alarmism is not warranted. Overall, educational uses of AI, such as personalized tutoring or teacher support, have enormous potential, and in the long run they can improve education and mental health too if designed well. But in the short term the transition demands clear rules: specific protections for minors, independent audits and informed consent from schools and families before anything is switched on. The yardstick should be evidence of safety, not speed of adoption.

The winners will be those who treat school deployment with the same rigor demanded of a medical product. The losers, those who use it as a lever for market share over a captive audience.