Poison Fountain and the rise of an underground resistance to AI

Poison Fountain, a shadowy group of technologists, has launched an underground resistance movement aimed at sabotaging AI systems by contaminating their training data with deliberately flawed code and text. The project encourages website operators to embed links pointing to "poisoned" content that AI web crawlers would collect, potentially degrading large language models during training. Recent research by Anthropic suggests this approach could be effective, finding that as few as 250 malicious documents can cause AI models to output gibberish.