Generative artificial intelligence (GenAI) tools are rapidly changing adult online learning environments. This systematic literature review synthesized 31 empirical studies on GenAI use in adult online learning and examined implementation contexts, reported outcomes, authors’ perspectives, and the roles attributed to GenAI. Drawing on peer-reviewed empirical studies published between 2020 and 2025, the findings indicated that most implementations embedded ChatGPT-based tools into online or hybrid courses for undergraduate, postgraduate, and preservice teacher populations. These tools positioned GenAI as a tutor, simulator, feedback provider, and course assistant. Outcomes are generally positive; the effectiveness of GenAI varied considerably, depending on the learning contexts, task designs, and model capabilities. A concerning trend was also detected, where GenAI tools were largely concentrated around models developed by OpenAI, reflecting limited diversity in the tools explored. The GenAI uses were generally correlated with improved learning performance, motivation, and self-efficacy, though concerns about accuracy, bias, and over-reliance persist in the reviewed literature. Overall, the findings suggest GenAI implementation is promising but cautionary, emphasizing the need for responsible and transparent integration.