Automated news summarization addresses the challenge of efficiently summarizing large amounts of information. In this digital age, people receive so much news content that they cannot afford to spend most of their time reading the entire article. Additionally, a summary can make news articles easier to understand for people who struggle with complex language or lengthy text, such as non-native speakers or those with cognitive problems. Automated text summarization techniques use machine learning algorithms to extract key concepts from lengthy articles, producing concise summaries that allow readers to absorb information quickly. This article proposed an approach for summarizing news implementing models XLNet and GPT-2 with other models such as BART, T5, Pegasus, LED, Big Bird, and Distill BERT on the BBC News dataset from the Hugging Face library. XLNet is a bidirectional model that analyses all words in a sequence to consider interdependencies, whereas GPT-2 generates diverse and coherent text based on context learned from a large corpus of data. The results suggest that GPT-2 performed better due to its capacity to generate coherent and contextually appropriate text across various categories and tasks. These findings highlight the importance of advanced machine learning models to improve the accuracy and efficiency of news summarization systems. Our research adds significant knowledge to the field of news summarization, for news editors looking to improve their summarizing methods.