<p>Astroturfing, a deceptive strategy designed to mimic grassroots movements, undermines the authenticity of online discourse and threatens democratic integrity. By orchestrating coordinated inauthentic behaviour, such campaigns distort public opinion and influence decision-making processes during critical events such as elections. Although detection methodologies have advanced, the increasing sophistication and dynamic coordination of astroturfing networks continue to challenge existing approaches. To address these gaps, this research introduces HiLo-TurfGaze, a hybrid framework that integrates unimodal content analysis with advanced clustering algorithms. Using the Global Political Tweets dataset, the model captures sentiment dynamics and hashtag co-occurrence to identify coordinated groups. The framework employs Louvain for network-level detection and agglomerative hierarchical clustering for refined community delineation. Comparative evaluation against established methods—including K-means, spectral, agglomerative, GMM, HDBSCAN, and Leiden—demonstrates the superiority of HiLo-TurfGaze. The model achieves the highest Silhouette Index (0.941), the lowest Davies–Bouldin Index (0.493), and the strongest Calinski–Harabasz Index (91,132), reflecting highly separated and cohesive clusters. Furthermore, HiLo-TurfGaze records the highest Homogeneity (0.219), Adjusted Rand Index (0.412), and normalized mutual information (0.487), indicating strong alignment with underlying data structures. While its execution time (~ 4 min) is moderately higher than that of simpler methods such as K-means (2 m 25 s), the substantial gains in clustering validity and interpretability outweigh this cost. These results underscore HiLo-TurfGaze as a robust benchmark for astroturfing detection, offering a scalable and reliable approach to uncovering coordinated disinformation. Future extensions will focus on multimodal data streams for real-time monitoring and enhanced resilience against emerging manipulation tactics.</p>