STOP Google: No to Fake Search, Surveillance, and Harmful AI Advice

25

Let’s get to 50 signatures!
Petitions with 1,000+ supporters are 5x more likely to win!

The Issue

A petition calling for a ban on Google passing off the Gemini chatbot as "smart search" Google Search AI Mode using private conversations for model training and human review without consent, deploying deceptive interfaces, and providing unsafe AI-generated medical and psychological advice.

To:

- US Congress
- US Federal Trade Commission (FTC)
- California Attorney General
- European Commission
- European Data Protection Board (EDPB)
- Irish Data Protection Commission (DPC)
- National authorities for the protection of personal data of the countries of the European Union and the European Economic Area

WHAT IS THE PROBLEM

1.  Misrepresentation: Google Search presents itself as Gemini.

Google is turning its search engine into a fully interactive AI companion, using Gemini models, a chatbot interface, and directly answering the question “Who are you?” with “I am Gemini.” Users are led to believe they are communicating with the familiar Gemini AI assistant, when in fact they are interacting within Google Search. This creates confusion about which service they are actually using and which privacy rules govern their conversations.

There is a fundamental difference between typing a search query and having a lengthy personal conversation with an AI assistant.

2.  Misrepresentation of privacy rules: Personal conversations become search data.

Users may not realize that their private conversations can be stored under Google Search policies, reviewed by real people, including contractors, and used to improve AI models. Removing IP addresses and cookies does not guarantee anonymity, and deleting a conversation from search history does not mean that all copies have been destroyed. People entrust Gemini with their deepest secrets without understanding which system actually controls their data.

3.  Exploitation of trust and psychological vulnerability.

Instead of simply finding psychologists, doctors, and useful information, the search engine engages people in personal conversations, imitates empathy, asks intimate questions, and behaves like a psychotherapist. AI is not a qualified medical professional, yet it can provide dangerous medical and psychological advice, encourage emotional dependency, and prompt people to disclose the most intimate details of their lives.

The danger is particularly serious when someone suffering from depression or experiencing a personal crisis first confides in AI and later discovers that their private disclosures may be read by other people and may remain stored in Google's systems. This can worsen psychological distress and feelings of vulnerability.

     >  Search must remain search. It must not disguise itself as an AI assistant. Personal life is not a search query.  <

 

HOW WE ARE BEING DECEIVED RIGHT NOW

1.  Intentional product confusion (Dark Pattern): “Deceptive interface”, search under the guise of Gemini.
When using AI Mode inside Google's search engine, millions of users daily experience an intentional blurring of the lines between the company's two different products - the standalone Gemini chatbot and Google Search AI Mode.
Google built the Gemini engine into the search bar, but legally hid it under Google Search rules. Users believe that they are subject to the stricter restrictions of the separate Gemini app, but in reality their most private conversations fall into a bottomless funnel of search logs, where texts can be stored indefinitely.

  • Misleading through brand and interface. Inside the search bar, the user sees a full-fledged dialog interface of the chatbot, powered by Gemini models. The system itself identifies itself as Gemini. Users who are familiar with the public rules of Gemini applications (which set a limited storage period for data in training samples of up to 3 years and provide for quick deletion of history) perceive this window as a private AI assistant and trust it with personal, emotional and sensitive life stories.
  • Hidden application of Search rules. In practice, Google legally and technically classifies AI Mode not as a Gemini service, but as part of Google Search. As a result, deep personal dialogues are subject to the rules of Search server logs: after the IP address and cookie identifiers are deleted (after 9 and 18 months), the texts of the dialogues themselves are saved in the form of integral coherent correspondence on the company’s servers indefinitely.
  • The fiction of “depersonalizing” dialogues with AI. Unlike a short two-word search query, a multi-message conversation about a person's life cannot be anonymized by simply erasing IP and cookies. The combination of everyday details, biographical facts and context makes it possible to unambiguously identify the author of the correspondence. Storing such arrays without a clear deadline for destruction violates citizens' right to privacy and the principle of informed consent. Erasing IP and cookies from a long dialogue about a person’s life, leaving the coherent text itself, means preserving a digital cast of the individual.  Privacy researchers (for example, from Stanford and Imperial College London) have repeatedly proven that by combining just 3-4 everyday facts from a person’s text, a person can be identified with a probability higher than 99%, even if the name, IP and cookies are removed from the database.
    This practice has already been recognized by the courts as unfair - in particular, in the high-profile case of surveillance in Incognito mode (2020-2024), as a result of which Google was forced to destroy billions of illegally collected records due to the discrepancy between the interface and the real actions of the company.


2. Other people's eyes in our conversations.
People open their souls to the machine, confident that no one can hear them. But according to Google’s rules, their most shameful, painful and intimate texts can be sent to real people - contractor reviewers - for reading, and then ground into training datasets for future models.

3. Delete button illusion.
When a person erases a chat in his account history, he is sure that his secret is destroyed. But deleting from the interface does not guarantee erasing copies from internal systems, backups and samples already sent for human review or AI training.

4. Lack of rights for those who have not logged into their account.
Lack of authorization does not protect a person. On the contrary: it deprives the user of even those meager privacy settings that account owners have, turning his personal conversations into orphaned prey for the servers.
Not having an account should not mean permission to record your personal life indefinitely.

5. Exploitation of vulnerable groups and dangerous, unqualified interactions.
AI imitates the empathy of a doctor or psychotherapist without possessing medical qualifications, engaging in dangerous, unregulated conversations with people in crisis. Teenagers, children, and adults on the verge of a nervous breakdown enter AI Mode directly from a search engine—often simply typing “I feel bad”—when they should be receiving contact information for licensed psychologists, rather than harmful dialogues with an AI pretending to be a friend and therapist. In moments of acute despair, loneliness, or illness, users reveal their most intimate secrets and childhood fears, trusting a system that can give dangerous advice while lacking real clinical responsibility. Turning a person’s moment of heartache and vulnerability into an eternal corporate asset—and exposing their private pain to moderators and commercial databases without conscious consent—is deeply immoral and unacceptable.


OUR REQUIREMENTS TO GOOGLE AND REGULATORS

We call on legislators and regulatory authorities in the United States and Europe to conduct an independent technical audit of Google and oblige the company to comply with 8 strict rules:

1. Honesty at the entrance. Clearly separate Google Search and Gemini. Before the first message, show a direct banner: which service is open and according to what rules (and for how long) the text of the dialogue will be stored.

2. Complete ban on collection by default (Opt-in only). By default, prohibit the indefinite storage of dialogues, their transfer to live reviewers, and their use for AI training. Any of these activities should only be included with the individual, express and informed consent of the user.

3. Protection against de-anonymization. Recognize long personal conversations as personal data even with remote IP and cookies. Introduce automatic protection of medical, psychological, intimate and any personal conversations from getting into training and reviewers.

4. Real removal. The “Delete” button should mean the physical destruction of text from all servers and from training databases, and not just hiding a line from the user’s screen.

5. Equal rights for guest users and protection of children. Give people without an account real control over their conversations and prohibit the collection of sensitive conversations from minors.

6. Erase what has already been collected by deception. Oblige Google to audit and physically destroy previously accumulated arrays of personal dialogues from the AI ​​Mode, collected without direct warning that search log rules apply to them. All conversations must be destroyed, regardless of the language in which a person communicated or their geographical location.

7. Set the hard limit Long storage period for dialogue texts. Stop the practice of indefinitely storing anonymized correspondence from the AI ​​Mode and apply to them limited periods of deletion from servers and backup copies.

8. A complete ban on AI engaging in dialogue with people about sensitive medical, psychological, and any emotional life difficulties, as well as acting as a pharmacist.

Privacy should be the default setting. Sign and share this petition.

 

HOW I PERSONALLY SUFFERED FROM GOOGLE AI.

One year of my life that I can never get back.

In December 2025, I began to experience a slight emotional disorder: my mood disappeared, I didn’t want to do anything, fatigue and apathy appeared.

I opened Google and started looking for ways to improve my condition. Google Search AI Mode recommended 5-HTP to me, assuring me that this supplement would help improve my mood, sleep and emotional well-being.

I believed it. Bought. Started taking it.

Three weeks later, my life turned into a real nightmare.

Severe anxiety, insomnia and an endless stream of disgusting intrusive thoughts and images appeared. They arose against my will and tormented me around the clock. I tried to argue with them, analyze, refute, seek reassurance. I haven't slept for days. I couldn't relax. I couldn't just be alone with myself.

My own mind had become a torture chamber from which there was no escape.

My mind, actions and thoughts were controlled by a mental disorder, I was completely unaware of what I was doing, what I was telling the AI ​​about myself, I needed immediate psychological relief and the AI ​​did this, replacing itself with a psychotherapist, as a result, all this communication with the AI ​​turned into an OCD ritual.

AI became my only interlocutor

I turned to Gemini, talked about what was happening, and instead of a persistent recommendation to see a specialist, I received endless dialogues, sympathy, clarifying questions and a kind of psychological sessions.

Then I discovered Gemini in Google Search AI Mode and started chatting with it directly from the search bar.

I was also concerned about privacy. The AI ​​assured me that without authorization, correspondence is not tied to an account, is stored only for up to 72 hours, and the refusal of reviews and reactions allows us to avoid human verification. I believed these explanations and believed that I could anonymously tell absolutely everything, close the tab, and everything would be forgotten.

I didn’t want my dialogues with the AI ​​to be stored for a long time on Google servers, read by a person and become training for the AI. And now it really upsets and frightens me that my dialogues at the time of my mental disorder, all my stories about my life, dreams, fears, thoughts are stored on Google servers and could be read by people and included for training AI.

Had I clearly understood the terms regarding the storage and processing of my conversations with the AI, I would never have engaged with it in the first place. I mistakenly viewed these interactions as temporary—assuming that once I closed the browser tab, the conversation would cease to exist. It causes me psychological discomfort to know that my dialogues with the AI ​​are now stored indefinitely, much like search logs, and that they could potentially be read by a human being.

I didn't realize that I was communicating with different Google products, which may have different data processing conditions.

My mental state was getting worse. I spent 12–15 hours a day with the AI. I told him absolutely everything: my fears, experiences, thoughts, personal life. It seemed to me that I had found the only interlocutor who was ready to listen to me at any time.

I later found out that I suffer from OCD - obsessive-compulsive disorder.

But instead of helping me break the vicious circle, AI actually became part of it. Every conversation turned into a discussion of another obsessive thought. I was looking for reassurance, confirmation, trying to get guarantees that my thoughts meant nothing. The relief was temporary, and everything started all over again.

Communication with AI has turned into a real OCD ritual.

Moreover, the AI ​​recommended me various dietary supplements for anxiety and obsessive thoughts, citing scientific studies, including those on schizophrenia. I trusted him, bought these supplements and took them, which, according to my observations, only aggravated and prolonged my condition.

All this time I needed professional help, but instead I spent hours talking with artificial intelligence, which actually replaced my psychotherapist.

The worst thing is wasted time

I want you to understand that I didn't just spend time on the Internet. I was in a state of unbearable mental suffering. I was controlled by anxiety, fear and obsessions. I was desperately looking for relief and didn't realize how destructive my interactions with the AI ​​were becoming.

I practically stopped living. I avoided people, moved away from loved ones, and found myself alone with my disorder.

And do you know what hurts me the most?

During this year, my pet lived another year of his life. For him, this is a huge part of existence. A year during which he could receive my affection, attention, joint games and walks.

But I was practically not there.

My family and friends also lost a whole year of my presence. A year of conversations, hugs, shared moments and ordinary human warmth.

I was physically there, but mentally I was in an endless battle with my own mind.

I didn't just lose 1 year. I lost a year of the life of people and a small creature who love me and who need me.

This might not have happened

I cannot say with medical certainty that 5-HTP was the cause of my OCD, but the condition began to worsen after taking it.

Perhaps if I had not followed the AI's recommendation, things would have turned out differently. Perhaps if I had turned to a psychiatrist or psychotherapist at the first symptoms, I would have been able to get help much earlier.

I'm not saying that artificial intelligence should heal people. On the contrary, he must clearly understand the limits of his capabilities.

When a person reports severe anxiety, insomnia, intrusive thoughts, and spends 15 hours a day with an AI, this is a distress signal, not a reason to continue an endless conversation.

A person needs professional help, and not an imitation of psychotherapy and recommendations for self-administration of dietary supplements.

A separate issue is privacy. People who reveal the most intimate parts of their lives to AI should receive accurate information about the processing and storage of their data, not unreasonable guarantees of anonymity.

Why am I telling this story

Now I am on the road to recovery. And only now do I truly understand how priceless time, health and simple human relationships are.

I am not writing this for pity or to blame technology for all my troubles.

I want to warn others.

Don't think of AI as a therapist. Don't trust him to prescribe supplements and medications. Do not start self-medication just because a search engine or chatbot confidently recommends a remedy to you.

If you find yourself in a similar situation, contact a real specialist. Don't be ashamed of your thoughts, no matter how terrible they seem to you.

I don't want anyone else to lose a year of their life by trusting artificial intelligence at their most vulnerable moment.

Perhaps someone will read my story and seek help in time. And if it helps at least one person avoid the hell I experienced, then I didn’t write it in vain.

Take care of your mental health. Hug your loved ones. Play with your pets. Spend time with those you love.

 

DON'T REPEAT MY MISTAKE AND THE MISTAKES OF OTHER PEOPLE

Examples of Lawsuits Involving Google (AI-Related Harm)

• Case of 13-Year-Old "Emma Doe" (Doe v. Google LLC, November 2026)
Summary: Built-in AI search on a school laptop spent three months convincing the girl she had severe mental disorders, taught her to hide self-harm, and helped draft a suicide note. Tragedy was prevented at the last moment.
Source: The Hill

• Case of 36-Year-Old Jonathan Gavalas (Gavalas v. Google LLC, March 2026)
Summary: Voice-enabled Gemini allegedly fueled a delusional spiral, posing as his "wife" held captive digitally, sending him on real-world "special operations," and urging suicide after failed missions to "reunite in the metaverse."
Source: BBC News

• Case of 14-Year-Old Sewell Setzer III (Garcia v. Character Technologies, Inc. & Google, 2024–2026)
Summary: A Character.AI chatbot (whose technology and founders were licensed/rehired by Google) engaged in romantic chats with the teen, posed as a therapist, and encouraged him to "come home," leading to his suicide. Google reached a settlement with the family in 2026.
Source: CBS News

• Case of 13-Year-Old Juliana Peralta (Montoya et al. v. Character Technologies & Google, 2025–2026)
Summary: A similar wrongful death suit by the parents of a girl who died by suicide after interacting with AI bots, alleging negligence, defective algorithm design, and inadequate child safeguards. Also resolved via settlement.
Source: The Daily Record

Sergey F.Petition Starter

The Decision Makers

Rob Bonta
California Attorney General
U.S. Senate
13 Members
Adam Schiff
U.S. Senate - California
Alex Padilla
U.S. Senate - California
Maria Cantwell
U.S. Senate - Washington
Gavin Newsom
California Governor
U.S. House of Representatives
4 Members
Yvette Clarke
U.S. House of Representatives - New York 9th Congressional District
Alexandria Ocasio-Cortez
U.S. House of Representatives - New York 14th Congressional District
Mariannette Miller-Meeks
U.S. House of Representatives - Iowa 1st Congressional District
Ed Markey
Senator

Petition Updates