<p>Twitter is the most widely used social network to reflect the opinions and emotions of an individual publicly. In recent years, researchers have developed many Machine Learning based methods for detecting spam accounts on Twitter. However, spammers use deceptive behavior, which prevents the methods from working efficiently. In this paper, we present a novel technique that combines Negative Selection Algorithm (NSA) and Genetic Algorithm (GA). The Negative Selection Algorithm is one of the important pattern discrimination algorithms of Artificial Immune System. In the Classical Negative Selection Algorithm, the candidate detectors are randomly generated resulting in redundant and inefficient detectors that cannot cover the entire negative space. To overcome this problem, the NSA detectors are optimized by Genetic Algorithm, a bio-inspired algorithm that follows the DNA replication that occurs in our body. When GA is combined with NSA, the negative spatial coverage is maximized, resulting in better anomaly detection performance. These optimal NSA detectors are then used to build the classification model, which is further used to detect collective anomalies. The proposed method is evaluated using a Twitter dataset and shows better accuracy when compared to Random forest and SVM classification methods.</p>