This article provides a current overview of developments in artificial intelligence (AI) for PET imaging. In addition to an introduction to deep learning methods, it discusses applications such as image reconstruction and image segmentation. Furthermore, it reviews the latest literature on AI-assisted diagnostic, prognostic, and predictive models in oncology and neurology. Advances in vision-language models (VLMs) and large language models (LLMs) demonstrate potential for structured reporting and workflow optimization. Despite these promising developments, rigorous validation of AI models remains essential to ensure their generalizability and clinical applicability.