Anti-AI activists are promoting a data poisoning movement aimed at corrupting the training data of future AI systems like ChatGPT and Gemini, with the goal of making them less reliable. The article describes this as both an act of protest and a genuine security concern. Data poisoning is an actual area of AI security research, where attackers feed misleading or corrupted material into the raw data that large language models absorb from books, websites, and other sources. A poisoned model might answer a specific question incorrectly while appearing normal otherwise, or contain hidden backdoor codes triggered by specific phrases. Some activists flood the internet with misleading AI-generated content, while artists use tools like Nightshade to subtly alter images, making them harder for AI to learn from. However, poisoning is difficult because AI companies filter and clean datasets. The real worry for cybersecurity researchers is not ChatGPT getting a history fact wrong, but poisoned information affecting behind-the-scenes AI systems used by hospitals and banks.