Artificial intelligence is increasingly framed as a driver of sustainable entrepreneurship, yet the empirical literature remains young, fragmented, and largely unconsolidated. This review synthesises 84 Scopus-indexed studies published between 2019 and 2026, drawn from articles, reviews, and conference papers and assembled following PRISMA principles for identification, screening, and inclusion. Source quality was treated as an analytical variable during synthesis instead of a filter applied at the eligibility stage. Two analytical passes were combined: a bibliometric mapping of author-keyword co-occurrence using VOSviewer, and a systematic, inductively coded reading of the literature organised under five themes covering opportunity recognition, eco-innovation, resource optimisation and circular business models, organisational capability in small and medium-sized enterprises, and entrepreneurship education alongside the ethics of generative AI. A consistent pattern emerges across all five themes: artificial intelligence rarely acts on sustainable entrepreneurship directly. It works through knowledge management, green innovation, and dynamic capabilities, and its measured effect weakens sharply once these organisational conditions are missing. The co-occurrence map reinforces this reading: artificial intelligence sits as a hub connecting every theme without forming a self-contained research conversation of its own, which supports describing it as a conditional enabler and not a prime mover. The evidence base skews cross-sectional in design and concentrates heavily in Chinese and Gulf manufacturing samples. Southeast Asia, and Vietnam in particular, appears mainly through student-intention surveys, with firm-level or ecosystem evidence largely missing, marking the clearest opening for context-specific research. For research and policy in Southeast Asian economies pursuing state-led digital agendas, this points to a practical risk: investing in artificial intelligence without first building the absorptive capacity to translate it into sustainability outcomes may produce formal compliance instead of genuine transformation...