As part of its Digital Strategy, the EU introduced the first dedicated legal framework to regulate artificial intelligence—the AI Act. The AI Act entered into force on 1 August 2024, and the majority of its provisions will be effective after 24 months. The AI Act undertook the risk-based approach, thus categorizing AI systems, depending on their purpose, into different risk levels—unacceptable risk AI systems, high-risk AI systems, limited risk and minimal risk AI systems. In addition, the AI Act addresses general-purpose AI models (and AI systems based on them) separately. Each category of AI systems is required to comply with different set of requirements and restrictions. One of the objectives of the AI Act is to ensure a high level of protection of fundamental rights as enshrined in the Charter of Fundamental Rights of the European Union, including the right to non-discrimination. This chapter will assess whether the risk-based approach, undertaken by the EU legislator, will suffice to address possible algorithmic discrimination resulting from the use of AI systems, particularly those of high risk. The analysis is twofold. First, it will be examined whether the AI Act adequately identifies the targeted risk groups of AI systems for protection against algorithmic discrimination. Second, the requirements and restrictions that the AI Act imposes on the targeted risk groups will be examined. The findings suggest that the AI Act, with its risk-based approach, undoubtedly improves the protection against discrimination. However, it also leaves many issues unresolved. Thus, the AI Act should be perceived only as a starting point in providing a systemic legal response to the issues of algorithmic discrimination.