<![CDATA[Artificial intelligence (AI) is entering university governance, teaching, and language education across developing countries, yet ethical-leadership scholarship remains concentrated on well-resourced, digitally mature institutions. This paper reports a PRISMA-informed, PICO-guided literature review examining how ethical AI leadership can be theorized for multilingual and intercultural language education in developing-country higher education, with a conflict-affected higher-education setting in South Asia as an illustrative case. Fifty-six sources met inclusion criteria after a staged identification, screening, and eligibility process conducted by a single reviewer using open academic search engines and publisher platforms rather than subscription-indexed databases. Thematic synthesis identified four recurring barrier domains infrastructural access, institutional governance, socio-cultural capacity, and linguistic/data equity and three leadership theories (transformational, distributed, and ethical leadership) that jointly inform a proposed six-dimension conceptual framework for ethical AI leadership. The review finds that AI tools trained predominantly on high-resource languages risk marginalizing low-resource languages such as Dari and Pashto, and that governance models assuming dedicated oversight offices do not transfer to capacity-constrained institutions. The paper contributes a context-sensitive, evidence-informed framework and a transparent, reproducible search protocol for future empirical validation in developing-country and conflict-affected higher-education settings.]]>
