<![CDATA[Purpose: Artificial intelligence (AI) is increasingly promoted as a transformative tool for improving mathematics learning, yet its educational value remains contested when issues of access, teacher readiness, and pedagogical safeguards are considered. This study aims to critically examine whether AI genuinely improves mathematics learning in secondary education by synthesizing evidence on its opportunities, risks, and safeguards. Method: This study employed a Systematic Literature Review (SLR) following the PRISMA 2020 guidelines. A total of 40 articles were selected from an initial pool of 309 records indexed in ERIC, Web of Science, and Scopus between 2021 and 2025. The selected studies were analyzed using thematic synthesis through open, axial, and selective coding to identify recurring patterns related to AI-supported mathematics learning, implementation risks, and responsible pedagogical practices. Findings: The findings indicate that AI can support mathematics learning through adaptive personalization, intelligent tutoring systems, automated assessment, real-time feedback, and increased student motivation. Evidence from the reviewed studies suggests that intelligent tutoring systems are associated with improved learning performance, with a reported aggregate effect size of g = 0.86 in relevant meta-analytic evidence. However, the review also identifies substantial risks, including unequal digital access, insufficient teacher readiness, weak TPACK-based integration, superficial learning, and dependence on generative AI. One reviewed study reported that students using generative AI without adequate guardrails performed 17% worse after AI access was removed, highlighting the need for responsible implementation. Significance: This review concludes that AI does not automatically improve mathematics learning. Its effectiveness depends on pedagogical safeguards, teacher mediation, equitable infrastructure, and blended instructional design. The study contributes a critical evidence-based framework for integrating AI responsibly in secondary mathematics education and provides practical guidance for teachers, school leaders, policymakers, and educational technology developers.]]>
