This systematic review examines research published between 2021 and 2025 on generative AI and chatbot use in secondary education across nine countries in Southern Europe and the Balkans: Greece, Italy, Spain, Portugal, Malta, Serbia, Croatia, Bulgaria, and Romania. Drawing on studies from IEEE Xplore, the ACM Digital Library, Google Scholar, and arXiv, this review synthesizes evidence on instructional uses, reported learning outcomes, teacher readiness, governance, and language-localization constraints. Across the region, the literature shows rapid experimentation in writing, language learning, programming, and project-based learning but limited long-term evaluation and weak cross-country comparability. Teacher interest is high, yet institutional guidance, assessment frameworks, and local-language resources remain uneven. This review argues that the next phase of adoption should move from isolated classroom experimentation to system-level implementation built around teacher AI literacy, transparent assessment, and context-sensitive design for smaller linguistic ecosystems.