<p>Artificial Intelligence (AI) is rapidly emerging as a pivotal technology in advancing Metal Additive Manufacturing (MAM)—a process that enables the fabrication of geometrically complex and customized metallic components beyond the capabilities of conventional manufacturing methods. While numerous studies and review papers have explored AI applications in specific MAM techniques—such as Laser Powder Bed Fusion (LPBF) or Wire Arc Additive Manufacturing (WAAM)—few have provided a comprehensive and balanced evaluation across the broader MAM landscape. Additionally, existing reviews often focus narrowly on individual processes or AI methods, lacking a unifying perspective to assess the field holistically. To address this gap, we introduce a novel, end-to-end AI-enhanced MAM lifecycle framework, covering the stages of Design, Build, Post-processing, and End-of-life, as the foundation for a systematic analysis. Using this framework, we move beyond descriptive summaries to deliver a critical synthesis of current progress, highlighting persistent challenges including data quality and availability, model interpretability, generalizability across materials and processes, and integration of domain knowledge. For each challenge, we highlight potential solutions, such as the use of physics-informed learning models and adaptive control frameworks. Finally, we propose a roadmap for future developments toward autonomous, intelligent, scalable and real-time MAM systems. This work serves as a foundation for researchers seeking to advance the integration of AI into MAM.</p> Graphic abstract <p></p>