LLMs that are trained by scraping the Internet will inevitably scrape the output of LLMs that have been used to generate content. The utility and trustworthiness of online sources will therefore decline over time and at an ever-increasing rate. This is the Ouroboros Syndrome.