Despite the progress in medical science, breast cancer has not ceased to be the major cause of female mortality all over the world. The urgency of the situation requires the creation of new, accurate and early diagnostic approaches. This work provides a completely new method for the detection of breast cancer using the YOLOv8 segmentation model. The advanced object detection capabilities of the YOLOv8 model were used to find cancerous parts in medical images. The particular feature of our model is that it was trained using a very comprehensive dataset of annotated breast cancer images, different data augmentation policies were used to strengthen the model’s robustness. This study presents a breast cancer detection approach using YOLOv8, which achieves 96% precision with high recall and IoU in the CBIS-DDSM data set. Comparing our detection strategy with other existing models highlights the advantages of the YOLOv8 model’s speed and accuracy. This confirms that the model is appropriate for clinical purposes, in addition to being able to perform the analysis instantly. This means that the YOLOv8 model we developed will be of great help in the timely detection of breast cancer symptoms and, thus, would lead to better patient outcomes if early intervention was followed instead of long-term treatment.