During drilling operations, abnormal pore pressures pose significant risks including blowouts, wellbore collapse, and pipe sticking, emphasizing the need for accurate pore pressure estimation in geoscience exploration and drilling projects. However, current prediction methods, whether mathematical or statistical, have limitations. Manual mathematical approaches are prone to human error, while statistical methods excel in inferring variable relationships but fall short in predictive accuracy. In response, artificial intelligence (AI), particularly deep learning (DL), offers promising solutions. This study proposes an innovative framework leveraging AI and DL for efficient pore pressure prediction solely from 3D seismic data. The framework includes end-to-end strategies, from preprocessing seismic data to predicting pressure zones. Data sourced from the Roosevelt, Utah field comprising 375 traces is employed, labelled using Eaton's pore pressure calculation and Gardner�s bulk density approximation. A DL model based on a simple deep neural network is integrated into the framework, incorporating techniques like training set resampling and hyperparameter optimization to enhance performance. Comparative analysis with state-of-the-art machine learning models validates the superiority of the proposed DL model. Results demonstrate its state-of-the-art performance, achieving 95.5 accuracy, 96.6 precision, 94.2 recall, and an F1-score of 95.3 in predicting pressure zones. This research showcases the potential of AI and DL in revolutionizing pore pressure prediction, offering advanced and cost-effective solutions to mitigate drilling-related risks in the oil and gas industry. © King Fahd University of Petroleum & Minerals 2024.