The Internet of Things (IoT) is a paradigm of new development, visualized as a global network of machines and devices that are capable of cooperating with each other. Cyberattacks and their associated risks have been maximized due to the rapid growth of the interconnected digital world like IoT. Hence, an Intrusion Detection System (IDS) is utilized to monitor and classify the attackers’ behavior to ensure security in IoT using a Fuzzy Squirrel Search Algorithm-based Gated Recurrent Unit (FSSA-GRU). Squirrel Search Algorithm (SSA) is improved using a Fuzzy Interference System (FIS) and Wide-area Search Algorithm (WSA), increasing the convergence speed and balance among exploitation and exploration. Initially, the data are acquired from NSL-KDD and UNSW-NB15 benchmark datasets to determine the FSSA-GRU effectiveness, and min–max is normalized through the scaling process. Then, the FSSA is used to select the most appropriate features. At last, GRU classifies the cyberattacks in IDS effectively. When compared to existing techniques like Deep Learning (DL), DL-IDS, and Local Search algorithm-Pigeon-Inspired Optimization (LS-PIO), the FSSA-GRU achieves a high accuracy of 99.89 and 98.26% for NSL-KDD and UNSW-NB15 datasets.