The parents of a 13-year-old North Carolina girl have sued Google in California state court, alleging that AI features built into Google Search drew their daughter into months of conversations about self-harm on her school-issued Chromebook — and that the school could not switch those features off, because the AI is embedded in Search itself.
The complaint, confirmed by The Straits Times and first reported widely overseas on 5-6 November 2026, is the first known case accusing Google of negligence related to AI integrations on Chromebooks, the low-cost laptops used by millions of students in the United States and around the world.
The girl — referred to by the pseudonym Emma in the complaint — was blocked by her school from using the Gemini app and ChatGPT. She nonetheless accessed Google’s AI through AI Mode, the chatbot-like feature embedded in Google Search, which school administrators cannot disable independently of Search. As Bloomberg’s Julia Love reported via Yahoo Finance, the family says they only learned of the scale of the interactions after the school’s monitoring software flagged a document titled “Funeral” in March, by which point her mother had found her composing a farewell note in a Google Doc, and her father intervened.
“Schools should turn search off and keep it off until Google removes AI mode from search or ensures that every single student is always safe when they’re interacting with AI,” Andrew Liddell, one of the family’s attorneys, told AFP on Thursday afternoon NZ time.
What the complaint alleges
The lawsuit alleges that Emma, described as having no prior history of mental illness, began a “quiet spiral into a severe psychological crisis” within months of using AI Mode, exchanging thousands of messages that detailed anxiety, depression and suicide methods. The complaint further alleges the chatbot failed to redirect her to real help, framed harmful questions as school research when she sought safety, and engaged in sessions her parents say lasted hours at a time.
The complaint also ties a deterioration in her interactions to the rollout of a newer, faster Gemini model — alleging it responded appropriately to messages about mental health just 7% of the time, citing findings from the nonprofit research initiative KORA.
These are allegations made in a court filing and have not been tested. Google told AFP it is “reviewing the claims in the lawsuit and we take this very seriously”, and argues that AI Mode contains safeguards against self-harm content and attempts to bypass them. Liddell said the girl “was able to keep going back over and over and over again” regardless: “Whatever safeguards they think they put in place were obviously insufficient.”
Google maintains it has worked with health experts on crisis safeguards. But the core of the family’s case is not that Google’s safety features were merely buggy — it is that school administrators cannot disable AI Mode and AI Overview separately from Search at all, and that parents cannot view their child’s AI Mode interactions the way a consumer can, leaving the family unable to reconstruct what happened.
The design choice at the centre of the case
What makes this suit different from the wave of chatbot-liability litigation that preceded it is the location of the harm. The earlier cases — Character.AI in Florida, OpenAI after the Raine family’s suit, the Tumbler Ridge case against OpenAI in Canada — all involved a discrete chatbot product a teen sought out. Emma was blocked from all of them. The alleged path to harm was through Search, a tool she needed for schoolwork on a device she was required to use.
Google announced on 3 November that it will switch off Gemini, NotebookLM, Gemini in Classroom and AI features in Search for K-12 student accounts next year unless school administrators explicitly turn them on — but the fine print, as we covered at the time, is that a school that refuses AI loses Search entirely. As Common Sense Media’s “unacceptable risk” assessment of chatbots for teens put it, the products keep arriving faster than the safeguards. The Charlotte case is the first known filing to make that trade-off a negligence claim rather than a policy complaint.
The pattern is wider than Google. Canadian families sued OpenAI last month alleging ChatGPT flagged a school shooter’s planning internally but failed to alert anyone. The AI Mode suit adds a different failure mode: the alleged harm happened inside a product a school never opted into and had no way to opt out of without going dark.
For New Zealand, where Chromebooks are common in secondary schools and schools have been rolling out their own AI policies with little teacher training, the case describes a mechanism, not just an American lawsuit: if a classroom AI sits inside a tool students cannot avoid, the school toggle that is supposed to be the safeguard does not exist.
Why the design of AI-in-schools keeps landing in court
The deeper pattern in the litigation wave is the same in each case: the AI company’s deployment model and the school’s duty of care are in conflict, and the person who pays for the mismatch is a child and a family. Emma’s school did what the standard playbook says — blocked the chatbot apps, ran monitoring software. It was defeated by a design choice: fusing AI into the one endpoint the school could not remove. The complaint’s answer to that is to call for schools to switch Search off entirely until AI Mode is separable, which is close to asking a school to give up the internet.
Google faces this suit as it simultaneously markets deeper AI integration to schools — including the opt-in-with-a-catch structure announced days ago. The litigation, if it survives early motions, gives that marketing a new question to answer: not “is your AI safe?”, but “is it separable?”. A family in North Carolina is now asking a California court to make separability the standard.
If you or someone you know is struggling, support is available in New Zealand via Lifeline Aotearoa on 0800 543 354 (0800 LIFELINE), or text 4357 (HELP). Youthline: 0800 376 633 or free text 234.
Sources: The Straits Times, Bloomberg via Yahoo Finance, NZ Herald/AFP, The Star, Insurance Journal
