This systematic scoping review, in accordance with the PRISMA-ScR guidelines, analyzed 12 articles published between 2017 and 2024, mostly from the USA and China. The most commonly used artificial intelligence (AI) models have been based on CNN and ViT, with recent studies increasingly using LLM-based models such as SkinGPT and Gemini for interactive analysis. Skin diseases investigated mainly included melanoma, nevi, basal cell carcinoma, keratinocyte carcinoma and seborrheic keratosis. AI models achieve high diagnostic efficiency for common and well-documented skin problems, but their accuracy drops significantly for rare diseases with insufficient data representation. The studies used both private clinical data and publicly available sources such as ISIC and MoleMap. Most AI models require better clinical validation and adherence to regulatory, ethical and legal standards. Despite limitations, AI models can improve dermatopathology by increasing the accuracy of lesion classification, facilitating early detection, and reducing diagnostic stress for clinicians and patients.