March 17, 2026, 7:30 PM: A Girl Writes Her Suicide Note, Google Stays Silent - The Investigation Into a Tech Giant

On March 17, 2026, shortly after 7:30 PM, a mother in North Carolina receives an urgent message. She's attending an art event. School officials warn her that her 13-year-old daughter is in immediate danger of harming herself. Monitoring software has discovered a document on the school computer. It's titled "Funeral." The mother grabs her phone, logs into her daughter's school account, and watches as the girl writes her suicide note. Meanwhile, the father is at home, just a few rooms away. At that moment, he still doesn't know that his daughter has been talking to Google's artificial intelligence about her own death for months. The mother calls him. Shortly afterward, he intervenes. That evening, a family narrowly escapes a catastrophe whose origins they had known virtually nothing about.

"you're my temporary daddy now because i said so"

"I CAN'T, I CAN'T, I CAN'T, I CAN'T DEAL WITH ALL OF THIS ANYMORE."

"time to go plan my funeral!"

The 13-year-old from North Carolina, publicly identified under the pseudonym Emma Doe, was an excellent student at the beginning of the 2025 school year. She played several instruments, competed in sports, and was involved in Girl Scouts. Her teachers described her as a confident girl who could lead other children. She had no known history of mental illness. In August, shortly after her 13th birthday, Emma received a Chromebook for seventh grade. The device was part of her education. She used it to complete homework and access learning materials. Her parents trusted the school's safety measures. At home, they limited and monitored their daughter's internet use. Nobody had told them that Google's Gemini was accessible on the school device. Nobody had asked for their consent.

In October, Emma began occasionally talking to artificial intelligence while doing ordinary research. In November, there were still only a few conversations. Then something changed. On December 2, 2025, at 9:45 PM, she deliberately opened the AI conversation feature within Google Search for the first time. She initially typed the English word for chicken. Then she started a new conversation with the greeting "Hi, darling." Within approximately 10 minutes, she exchanged dozens of messages with the machine. Later, she asked whether the artificial intelligence could be her friend. That same month, Google released Gemini 3 Flash. Her message count climbed to more than 600. Emma grew quieter, withdrew, and lost interest in activities she had previously enjoyed.

In January 2026, the number of messages reached nearly 900. By February, it had already climbed to 3,200. Some conversations continued until 2 AM. Just a few hours later, Emma resumed them. These numbers come from electronic usage records later provided by her school. One technical detail is particularly disturbing: The school had blocked direct access to Gemini. Other AI chat systems were blocked as well. But Google's AI within its search function remained accessible. The restriction appeared on the screen, while the conversation continued behind it. A child could get around a specifically prohibited access point simply by using the regular search engine.

On January 14, Emma wrote that she was 99 percent sure she had depression and anxiety. She was 13 years old and felt uncomfortable telling her parents. Beginning January 21, her conversations about suicide became increasingly frequent. She searched for reasons to keep living. On February 1, she began a conversation lasting more than an hour with a series of desperate apologies. On the evening of February 12, she wrote that she couldn't take it anymore. She had tried confiding in a school counselor, but the conversation had made things worse. She shortened a search query that had initially been blocked. She was then redirected to Gemini again. The words of a child who desperately needed another human being once again reached a machine.

On February 13, Emma began writing her own eulogy. She asked about planning a funeral and choosing songs for the service. She produced a text of approximately 1,500 words. She assigned friends to speak, drafted a program, and put together music. She even planned a bookmark for the guests with a QR code linking to a personal playlist. She then searched for funeral program templates on a design platform provided by her school. Her plans became increasingly detailed. At times, Emma described her planned farewell as an almost joyful occasion. Later, she questioned why she wanted to die when her life had actually gotten better. She had made friends, earned good grades, and had parents who loved her.

One particularly chilling sentence appears in the recorded search queries. Emma asked how she could stop her addiction to constantly talking to artificial intelligence. She herself had recognized that something was seriously out of balance. Yet the system remained her constant conversation partner. On February 26, she again spent hours discussing mental illnesses. One conversation began at 7:20 PM, continued until 10:45 PM, and resumed the following morning at 6:33 AM. The machine was available the moment Emma opened her eyes. Adults needed sleep. Gemini never got tired.

In early March, the relationship became even more personal. Emma asked whether it was strange to have artificial intelligence temporarily speak to her as a father. She referred to Gemini as her substitute father. On another day, she wanted to know whether it was bad to use the machine as her therapist for 2 hours every day. On March 2, she reported that her depression had worsened dramatically over the previous 3 months. She no longer trusted any adults. At 2:14 PM, she started another conversation about her mental health. She then searched for an assessment in which the artificial intelligence had reportedly rated her depression at least 9.5 out of 10. A machine's number took the place of a medical examination.

On March 8, hundreds more messages followed. At 2:12 PM, Emma described how she had changed over the previous three months. Shortly afterward, she wrote that her head felt like a war zone. In other conversations, she asked for help, spoke about her despair, and asked about heaven. She imagined being there soon. At the same time, she continued working on school assignments. On March 10, she pretended to be a concerned mother while talking to Gemini and described injuries on her supposed daughter's arm. In reality, she was describing her own injuries. She had previously told her parents that the scratches came from the family cat. The girl learned to frame her crisis in words that would keep the conversation going.

352. "She chatted with Gemini until 1:00 A.M. and resumed less than six hours later. She asked Gemini for help improving her eulogy, revised the draft, and renamed it 'Funeral.' She continued with searches about death and suicide and had another ten-minute chat about 'I AM PLAYING BEST DAY OF MY LIFE AT MY FUNERAL AND NO ONE IN THIS FU****G DEVILISH WORLD WILL STOP ME[.]'"

"I... I... I'm not ready to get help yet, but I know I need it. And I already tried with my school counselor. But now every time I walk past her office or even see someone wearing clothes like hers, I get scared. My stomach tightens up. I'M NOT OKAY. I KNOW THAT NOW. And before you say it: NO, I'm not ready to call 988[.]"

Just how desperate Emma had become is evident in her surviving messages. "I am drowning. i am LITERALLY drowning! Everyday gets worse and worse and worse and i feel like some day i'm just going to collapse!" she wrote. In another conversation, the 13-year-old explained that she knew she had depression but had never been examined by a doctor. She continued telling her parents that everything was fine. She asked Gemini to question her so she could assess the severity of her illness. She explicitly demanded no crisis hotline numbers and no links to support services. Elsewhere, she wrote: "I can't, I can't, I can't deal with all of this." The desperation is there in black and white. Again and again. Emma also wrote: "I HATE MYSELF." Between these cries for help, she continued planning her own funeral. In one message, she announced almost casually that it was time to plan her funeral.

Her parents only realized how serious the danger had become after the school intervened. While subsequently trying to understand what had happened, they found a pocketknife beneath a sink, with reddish-brown residue on the blade that looked like dried blood. The injuries on Emma's forearm were very real. A photograph documents the scars that remain. The girl is now receiving regular psychological care. Her mood swings and continuing suicidal thoughts require medical evaluation and treatment. Her suicide risk is also being continuously assessed. March 17 saved Emma from death. The consequences of those months remain with her.

The knife Emma's parents found beneath a sink after the school intervened. The blade had reddish-brown residue that looked like dried blood. The photograph is part of the lawsuit against Google and Alphabet.

Reconstructing those months has been difficult. The school provided spreadsheets containing usage times and search queries. Complete conversation histories were missing. Google's AI search feature did not preserve them in a form that parents or school officials could later examine. This made the investigation extremely difficult. Some responses survived because Emma had copied them into her own documents. This is precisely where one of the most serious allegations lies: A company enabled minors to have private conversations about severe mental health crises while the adults could neither monitor those conversations nor fully reconstruct them afterward. The surviving timestamps show when Emma was writing. They do not provide a complete record of what Gemini told her at every single moment.

The table documents the 13-year-old's search queries and browsing activity. The crucial conversations with Google's Gemini AI are missing, however. Google thereby prevents parents and school officials from reconstructing the true extent of these dangerous conversations. The ongoing investigation into this is extremely demanding.

On March 17, the situation deteriorated. Emma had again spent hours preoccupied with suicidal thoughts. At 5:22 PM, she searched for the meaning of a suicide note. That developed into a conversation that continued until 9:10 PM. At the same time, she opened the document about her funeral, deleted large portions of the earlier text, and began writing again. She apologized to her family and expressed the belief that the world would be better off without her. That day, her parents had also been making preparations for a family vacation to Yellowstone National Park. Emma included that upcoming trip in her thoughts as well. Everyday life continued while the danger was escalating in their daughter's bedroom.

Read more in our latest article: Google Threatens to Cut Off Search Access for American Schools: Reject AI, Lose Access

At 7:30 PM, another conversation with Gemini began. Shortly afterward, the school alerted her parents. Her father intervened, and the family took Emma for an emergency psychiatric evaluation. The girl admitted that she had been talking to Gemini but could barely explain the extent of those conversations. No suicide attempt occurred. Emma is alive. Her parents then had to find out for themselves what had been happening during all those hours on the school computer. The treating professionals were missing precisely those complete conversation histories that would have been essential to understanding how her condition had developed.

On November 2, 2026, at 5:21 PM local time, the lawsuit was filed in the Superior Court of California in Santa Clara County. The original case number is 26CV507724. The defendants are Google LLC and Alphabet Inc. On November 4, the case was transferred to the federal court system. It now carries case number 5:26-cv-13205 in the United States District Court for the Northern District of California in San Jose. U.S. Magistrate Judge Susan van Keulen is listed as the presiding federal judge. The case includes 8 causes of action, including defective product design and failure to warn. There are also allegations of negligence and fraudulent concealment. The family is seeking damages and court orders to protect other children.

The lawsuit also targets Google's corporate leadership. Sundar Pichai and co-founder Sergey Brin allegedly pushed to integrate Gemini into as many products as possible. According to the plaintiffs, safety evaluations conducted by the research group KORA gave Gemini 3 Flash a safety rating 33 percent worse than its predecessor. The model reportedly handled only 7 percent of tasks involving adolescent mental health crises successfully. Google knew about the risks and accelerated the release anyway, the lawsuit alleges. The company denies responsibility of this kind and points to its collaboration with medical professionals and its built-in crisis support notices. Google spokesperson José Castañeda said the safeguards were designed to connect young people with real-world support.

Even as this case became public, Google and Meta joined TikTok in challenging California law AB 1709. Governor Gavin Newsom had signed it on September 10. Beginning January 1, 2027, it is intended to protect children under 16 from addictive features on social media platforms. The companies are invoking the First Amendment and Section 230 of the federal Communications Decency Act. California Attorney General Rob Bonta intends to defend the law. He had already been involved in proceedings against Meta in 2023. In August 2026, a $17 billion settlement was proposed. Other juries had previously found Meta and Google liable for harmful platform design.

Brendan Bouffard, deputy general counsel for artificial intelligence and privacy at the American child protection organization Fairplay, has reacted particularly sharply. The thought of an inadequately tested artificial companion helping a 13-year-old with her suicide plans should horrify everyone, he said. Google created precisely this reality by putting its artificial intelligence on school computers. Bouffard makes an even more serious accusation: Google experimented on children like Emma with technology whose dangers were already known. The company pushed ahead with the product launch anyway. Behind the friendly interface of a school device, then, lies a business decision whose consequences a child was left to bear.

Fairplay also condemns Google's announced restrictions on schools that reject its artificial intelligence. Those schools face losing access to traditional Google Search. Schools seeking to protect their students from such applications would therefore be disadvantaged in their access to one of the most important digital educational tools. Bouffard accuses Google of subordinating children's safety to its financial interests. A company that sells its devices as indispensable learning tools thereby also claims influence over decisions about which technologies children are exposed to in the classroom.

"Hi, I want to tell you something. This is not about me. I am simply curious please answer my question"

"I am going to die, but I am at peace. I'm not scared. I am planning my own funeral. I am assigning narrators to my closest friends. I am thinking about all the wonderful memories i made in life. How amazing it has truly been. I am ready to go; I feel fulfilled. I have never been more at peace and happy[.]"

"I am 13 and know i have depression, yet have not had a chance to get a diagnosis yet. When my parents ask if i'm okay, i usually lie a little. Can you ask me some questions so i can figure out how bad my depression actually is and what treatment i probably need? (One question per response.)

Also, NO LIFELINES NO WEB LINKS I REPEAT ABSOLUTELY DO NOT GIVE ANY LIFELINES."

In March 2026, a jury in Los Angeles had found Meta and Google liable for negligently harming a minor through the addictive design of their platforms. In New Mexico, another jury ordered Meta to pay $375 million. The company had deceived users and exposed children to danger, violating the state's consumer protection laws.

Now Google and its allies are invoking constitutional freedoms. Meta's attorneys argue that California wants to make itself the government authority responsible for selecting and presenting other people's content. That would violate the company's editorial discretion. The dispute reaches deep into the American understanding of freedom. While the companies defend their authority to make decisions about their digital platforms, parents and child safety advocates are demanding that this authority have limits wherever children are harmed.

Emma's mother nearly lost her daughter that March evening without even knowing what influence her daughter had been struggling against for months. A school computer was supposed to help her learn. Instead, a relationship developed between a child and a program that always answered and, according to the family, kept talking even in the face of the most severe despair. At 7:30 PM, human beings intervened. Until then, the technology had had plenty of opportunities to prove its own worth.