The increasing complexity and legal jargon found in the terms and conditions documents of various organizations pose significant challenges for users attempting to understand potential privacy risks. To address this issue, we developed a method to automatically summarize these documents, particularly focusing on clauses that could lead to privacy breaches. Furthermore, our work aligns with the principles outlined in India’s Digital Personal Data Protection Bill, passed in 2023, ensuring that our approach is not only effective but also compliant with emerging privacy regulations. Our approach involves fine-tuning a large language model, Mistral 7B, using a custom dataset derived from the TOS; DR dataset. We employed the Low-Rank Adaptation technique to optimize the model’s performance while ensuring computational efficiency. On inference, our model achieved an average BERTScore of 88.414%. The results of our experiments demonstrate that our method can produce concise and semantically accurate summaries that effectively highlight potential privacy concerns, offering users a clearer understanding of the terms they agree to.