Social media platforms have emerged as powerful tools for fostering instantaneous connections and networking across geographical boundaries. However, it is imperative to recognize that these platforms also harbor potential detrimental outcomes. The proliferation of bots contributes to the propagation of unsolicited messages, fraudulent schemes, and a spectrum of cyber threats, thereby compromising users’ data security and privacy. Previous research pre-dominantly centered on crafting bot detection models tailored to specific social media platforms. This has led to a dearth of holistic endeavors aimed at creating a unified bot detection technique effective across diverse social media networks. This paper introduces a generalized approach, GRUbBD, designed to detect bot accounts on social media. The GRUbBD model leverages user profile metadata-derived attributes and behavioral patterns for bot detection. Integrating a Gated Recurrent Unit (GRU) and an attention mechanism, the GRUbBD model is engineered to bolster its performance. To streamline computational efficiency, a dimensionality reduction technique, specifically linear discriminant analysis, is applied during data preprocessing phase. The effectiveness of GRUbBD is demonstrated by empirical evaluations performed leveraging datasets sourced from Twitter and Instagram. The GRUbBD model has a noteworthy level of accuracy, achieving 99% and 92% accuracy for the Twitter and Instagram datasets, respectively. This research contributes to the advancement of bot detection strategies while offering broader applicability across diverse social media platforms.