Large Language Models (LLMs) have been the most recent breakthrough technology to understand and generate text in the manner that is human-like, revolutionizing industries from customer service to creative content generation and further advancing natural language processing across sector after sector. In simpler words, these have made human-computer interaction intuitive and efficient. This work uses LLMs to extract landslide-related information from online news articles. With increased digitization and availability of web-based news media, a huge number of news articles on landslide disasters are available across several countries in many languages. The research extracts metadata on the locations of such incidents from news articles about landslides occurring in India. The extracted locations and the corresponding details of the news article are geocoded into coordinates for further analysis. The study further presents an algorithm developed for accurately disambiguating the location mentions in the Indian context. A comparison with state-of-the-art methods shows that the proposed algorithm outperforms on a few particular cases. A case study on landslide events on May 2024 is also provided.