<p>To address the growing interest in, yet limited understanding of, ChatGPT-supported inquiry-based learning in STEAM education, this study systematically reviewed 24 empirical articles to identify current research trends, examine ChatGPT’s roles across the five inquiry phases, and analyze its advantages and challenges. The reviewed studies showed an uneven disciplinary distribution, with science receiving the most attention, whereas technology, the arts, engineering, and integrated STEAM remained underrepresented. The studies were evenly distributed between K–12 and higher education contexts; however, undergraduate students constituted the largest single participant group, and no study focused on elementary learners. ChatGPT was primarily used during the conceptualization, investigation, and discussion phases, where it functioned as a learning tool, tutor, learning peer, domain expert, and teaching assistant to support question formulation, inquiry design, problem-solving, and reflection. Its use in the orientation and conclusion phases remains largely unexplored. The studies reported several advantages of using ChatGPT. For students, it enhanced academic performance, critical thinking, engagement, and motivation, while also supporting individualized learning. For learning and teaching processes, it improved interaction, reduced educator workload, and facilitated differentiated learning environments, ultimately enhancing teaching efficiency. However, the studies also identified multifaceted challenges. At the learner level, students may over-rely on ChatGPT-generated answers, accept inaccurate or irrelevant feedback, experience conceptual confusion, or produce superficial conclusions when AI-generated outputs are treated as authoritative rather than critically evaluated. ChatGPT may also provide generic or insufficiently personalized responses that fail to stimulate divergent thinking. In addition, hallucinations, including inaccurate or fabricated explanations, may undermine evidence synthesis and reasoning during the conclusion phase. At the educator and institutional levels, challenges included misalignment with curricular goals, reduced instructional depth, the need for sustained teacher mediation, unequal access, technical barriers, privacy concerns, risks to academic integrity, and algorithmic bias. An integrated framework is proposed that synthesizes ChatGPT’s roles, advantages, and challenges across the five phases of inquiry-based learning, along with four agendas for future research. These findings offer guidance for responsible and pedagogically aligned integration of ChatGPT into STEAM inquiry-based learning.</p>