Large language models (LLMs) like ChatGPT are powerful tools, but they sometimes generate incorrect or fabricated information—a phenomenon known as "hallucination." How can we trust what we read from these AI systems? New research explores ways to help users identify these hallucinations and build appropriate trust. Researchers investigated different methods of showing users how factual an LLM's response is, including color-coding phrases based on factuality scores and highlighting sections of source material used to generate the response. The study found that users preferred color-coding, which made it easier to spot inaccuracies and increased trust in the AI. Showing source attributions, such as highlighting relevant source text or using reference numbers, also boosted user trust. However, some users found these additions overwhelming, suggesting a need for customizable interfaces. This research provides valuable insights into designing AI systems that promote transparency and help users navigate the complexities of AI-generated content. While these tools are promising, it's crucial to remember that even with factuality indicators, users should always cross-check information from multiple sources to avoid over-reliance on any single AI system.