CHATBOT HALLUCINATIONS ARE POISONING WEB SEARCH

Summary from the TLDR InfoSec newsletter

A study of how generative AI chatbot “hallucinations”, or false information generated by AI chatbots, are harming search results. The study focuses on an example in which a researcher wrote a blog post that was meant to test how an LLM could be tricked into generating a summary using a fictional article. The researcher discovered that their fictional article was highlighted in Bing search results even after they added explicit text in the blog post to try and dissuade the Bing search results.