Artificial Intelligence (AI) is increasingly embedded in the software development life cycle (SDLC), reshaping the design, implementation, and maintenance of software. However, the impacts of sustainability remain insufficiently explored. This study investigates how AI affects sustainability across SDLC phases by integrating insights from academic research and industrial practice. We applied a mixed-method approach combining a systematic review of 64 peer-reviewed studies with 27 semi-structured interviews conducted across 12 countries. The analysis covers five dimensions of sustainability: environmental, economic, social, individual, and technical, highlighting both positive “handprints” and negative “footprints.” The results indicated that AI is most widely adopted during implementation and testing, with limited use in the requirements, design, and deployment phases. AI improves productivity, code quality, and testing efficiency. However, major challenges persist, such as unmonitored energy consumption, vendor dependency, bias, skill degradation, and AI-related defects. Based on these findings, we propose actionable recommendations for integrating sustainability checkpoints, such as energy and cost monitoring, model size optimization, human oversight, and AI-aware quality reviews, into AI-assisted workflows. These measures aim to shift from incidental efficiency to intentional sustainability throughout the SDLC.