The advent of deepfakes, highly realistic synthetic media produced by advanced AI algorithms, has added another level of complexity to combating misinformation and digital deception. As generation algorithms advance, detection of such manipulated videos and images becomes challenging. Although deep learning has exhibited some capability in identifying deepfakes, such models have challenges pertaining to overfitting, generalization issues, and computational expenses. Swarm Intelligence (SI), drawing ideas from collective behavior of natural systems such as flocks of birds or colonies of ants, provides a new angle. By optimally and efficiently tuning model parameters, filtering out appropriate features, and enhancing ensemble approaches, SI algorithms are investigated to enhance detection accuracy and resilience. The propose survey integrates existing research where swarm- inspired approaches such as Particle Swarm Optimization, Ant Colony Optimization, and hybrid variants thereof are employed to detect deepfakes. Survey on their merits, shortcomings, and performance on diverse datasets and applications. Paper concludes with an overview of the main challenges at this boundary of fields and their promising avenues of future inquiry.