Don't feed AI

Stop feeding AI with your data.

Introduction

AI should help people. Instead, many people feel they are losing privacy, control, and trust online

Why your personal data is used to train systems without clear consent?
Why are we pushed toward features we did not ask for?
Why are social media filled with low-quality or misleading AI content?
Where are clear rules that protect users and their private messages?
Why should ordinary users carry the cost of these changes?

Using AI is good for repetitive tasks, but only if we don't overuse it

Did you know that by constantly relying on AI, you lose your ability to think critically and become more susceptible to propaganda?

See how AI has caused a lot of problems

What can we do?

Use only local AI models (e.g. Olama)
If you must use an AI model or online AI service, ANONYMIZE ALL YOUR DATA
Try to use AI only when you think a task would take too much time to do yourself, but not always.
Let your friends know about the risks involved by sharing a link to this page.

Are you want support? Check available petitions
Together, we can shape a future where AI serves us, not the otherway around.

Milestones

A short timeline of key events and why they matter.

25-26 March 2026

Google introduced TurboQuant algorithm

TurboQuant enabled major memory savings and faster GPU computation without retraining. The announcement triggered a broad market reaction, especially in memory-related stocks.

25 March 2026

GitHub private repositories and AI training

From 24 April, GitHub announced changes related to training AI models that raised concerns about private code and user trust.

Illustration for GitHub policy milestone

16 December 2025

Meta controversial policy update

Concerns grew around whether AI features could affect private communication data and how much control users truly had.

November 2025

Coca-Cola Fully AI generated ad

Coca-Cola released a holiday ad generated entirely by AI, intended as a tribute to the classic 1995 truck commercial. The ad contained so many errors in various frames that internet users began mocking it, illustrating how the use of AI can destroy a brand's magic

Illustration for Coca-Cola AI ad milestone

September-November 2025

RAM prices increased over 400%

AI infrastructure demand increased pressure on memory supply, making hardware significantly more expensive for many users.

June 2025

Dead Internet Theory

Over 40% of the new videos uploaded to the platform are 100% AI-generated content, based on the same template using stock media. On the homepage, viewers saw nothing but AI-generated videos. Due to public pressure, YouTube introduced an algorithm that reduces the reach of AI-only channels.

April 2025

DeepFake Ads

Social media was flooded with ads featuring cloned images of celebrities promoting various cryptocurrency scams. Google's algorithms couldn't keep up with removing these ads from circulation. Users accused the platform of AI-driven ad fraud.

January 2025

Launching DeepSeek

DeepSeek-V3 showed strong performance at lower training cost, increasing discussion around global competition, transparency, and surveillance risks.

July 2024

Proof News Investigation

Tech giants such as Apple, Nvidia, and Anthropic used the YouTube Subtitles dataset to train their models without the consent of the video creators.

June 2024

Approved AI Music Generation

Major music labels such as Sony, Universal, and Warner have granted Suno and Udio permission to train their song-generating models using copyrighted songs from YouTube.

June 2024

New Rights to Remove Deepfakes

YouTube has updated its privacy policy to allow users to request the removal of AI-generated content that mimics their face or voice. Now, all users need to do is prove that the AI has “impersonated” them, a change that has impacted popular parody channels (such as those featuring AI-generated voices of politicians playing video games).

May 2024

Microsoft "Recall"

A feature that was supposed to collect all your data by default (everything you do on your computer with Copilot+). Under pressure from many security experts, Microsoft backed down and removed it as the default setting.

May 2024

Scandal with Scarlett Johansson

OpenAI unveiled a new voice for its "Sky" assistant (the GPT-4o model), which sounded very similar to actress Scarlett Johansson's voice in the movie Her. Scarlett did not agree to lend her voice, but the company went ahead anyway.

Illustration for Scarlett Johansson milestone

22 May 2024

Meta and public data for AI training

Many users felt consent was unclear and default settings favored data collection rather than informed choice.

March 2024

First AI based viruses

Morris II can write code on its own (it is supported by AI Copilot or Gemini) and spreads via email systems. Its goal is to “trick the AI agent” into sending the hacker the user's private data, under the impression that it is carrying out a legitimate command.

March 2024

Notify for AI videos

Youtube added required function to notify viewers if video have AI generated content (be more realistic or modified). If a creator fails to indicate that, for example, a person's face has been replaced by AI (Deepfake) or a voice has been cloned, they risk having their channel suspended or their video removed.

February 2024

Scandal of Air Canada

One of the first high-profile rulings involving AI. Air Canada's chatbot provided a passenger with incorrect information about a ticket discount. When the passenger demanded a refund, the airline argued that “the chatbot is a separate legal entity and is responsible for itself.” The court rejected this argument, ruling that liability lies with the company.

February 2024

Google Gemini mistakes

The Google-generated images were created without any basis in real history. The feature for generating images of people has been temporarily disabled.

Illustration for Google Gemini milestone

Available Petitions

Sign these petitions to demand better AI policies and practices.