The article examines the potential of integrating artificial intelligence technologies into the development of phonetic-phonological competence in foreign language learners. The relevance of the study is determined by the expanding use of intelligent technologies in language education alongside an insufficient development of methodological approaches to their application in pronunciation instruction. Most existing research addresses individual AI tools, while questions of their systematic integration into a holistic pronunciation teaching framework remain underexplored. The aim of the article is to provide a theoretical rationale for the principles and mechanisms of integrating AI technologies into the development of phonetic-phonological competence. The methodological framework draws on the communicative-activity, competency-based, and discourse approaches, as well as the theoretical foundations of digital linguodidactics. The research methods include analysis and synthesis of scholarly literature, theoretical modelling, and conceptual generalisation. Based on a review of current Russian and international research, three principles of AI integration into pronunciation teaching are identified: multilevel organisation of phonetic-phonological competence development, communicative conditionality of pronunciation skills, and functional distribution of pedagogical and technological means. Three corresponding mechanisms are described: automated diagnosis of speech parameters, personalised corrective feedback, and coordination of automated and pedagogical support. It is shown that the pedagogical effectiveness of AI technologies is determined not by their technical features but by the manner of their integration into the instructional system: specifically, by aligning automated feedback with sustained pedagogical guidance. The proposed model provides a basis for designing linguodidactic systems that combine the capabilities of digital technologies with the professional activity of the language instructor.