Large Language Models have changed the way computers understand and use language. These models can create answer that make sense and fit the situation in areas. Even though they have improved they can still make up things that’re not true. These made-up parts can look real. They are not correct. This can make people not trust the information and can be dangerous in areas like health, money, school and law. This research introduces a method called Explainable Artificial Intelligence that helps find these made-up parts and explain why it thinks they are not correct. The methos uses the way to check if the information is right. It looks at how the words fits how the similar the meaning is, how sure the model is and how it makes decisions. Techniques like showing which parts of the text are most important and how the model decides help people understand why something might be wrong. The method was tested using data sets with false information with large model. The results were checked using ways to see how good it was. The study shows that the method work well at finding made-up information and helps people trust the system more. By making sure the model is correct and the way it makes decisions is clear this method helps build trustworthy AI system. It gives a start, foe work on using large models in a responsible way.