Offline handwritten signature verification is a biometric system that aims to verify individuals’ identities based on their unique writing style, commonly utilized in tasks involving documents such as bank checks and contracts. However, due to the reduced number of available signature samples for each writer (usually five signatures at most), such a system requires a significant amount of additional samples to achieve the desired performance. This challenge has been extensively addressed by a large research community over the last years, leading to the use of data augmentation in order to generate fully synthetic signature images. Presently, we investigate the data augmentation through the generation of synthetic signature features by using an artificial immune algorithm. Synthetic features are used to extend the signature training set of the verification system which is based on a SVM classifier.