Artificial intelligence (AI) is now embedded in many domains, yet persistent concerns remain about opaque decision processes, training-data bias, and impacts on privacy and human rights. The European Union’s AI Act addresses these risks through a risk-based framework that imposes stringent obligations on systems classified as high-risk. The Act’s voluminous and intricate provisions make it difficult for engineers and compliance teams to obtain an integrated view from the legal text alone. This paper employs Goal Structuring Notation (GSN) to translate the AI Act’s high-risk requirements into a layered argument model. The resulting GSN diagram exposes the logical links between individual articles, annexes, and the supporting evidence needed for conformity. The approach improves traceability, highlights documentation gaps, and offers a maintainable structure for tracking future amendments to the legislation. The resulting GSN diagram not only exposes the logical links between individual articles but also serves as a common ground for stakeholders to discuss and negotiate AI Act compliance.