Judicial systems worldwide, particularly in Global Majority countries, are witnessing a growing interest in artificial intelligence (AI) technologies, given their potential to enhance judicial efficiency and access to justice. The role of these technologies is no longer limited to automating administrative tasks or improving case management; rather, they have expanded to include assisting judges in analysing facts, identifying precedents, and even (albeit to varying degrees) influencing judicial decision-making. This study seeks to explore the integration of AI into judicial systems, with a particular focus on comparative experiences across Global Majority countries. It analyses both the opportunities and the legal and ethical challenges that such integration presents. The study builds on the findings of a recent UNESCO survey of judges from across the globe. The results indicate that while judges generally demonstrate awareness of AI, its professional application remains in early stages and is mostly confined to non-decisive tasks, with outputs carefully scrutinised before use. The paper proposes a functional typology of AI use in the judiciary, organised into three main categories: (1) assistive systems for administrative tasks such as automated transcription and file management; (2) decision-support systems based on precedent and information analysis; and (3) semi-autonomous systems that contribute substantively to judicial reasoning and judgment drafting. The study presents practical examples from countries such as Egypt, China, Turkey, the United Arab Emirates, Brazil, and Argentina. Conversely, the study highlights five key challenges that threaten the safe and responsible use of AI in judicial settings: (1) the risk of privacy violations through potential leaks or reproduction of sensitive data; (2) legal stasis due to an overemphasis on consistency at the expense of legal development; (3) erosion of judicial independence stemming from overreliance on AI recommendations; (4) opacity of justice resulting from the use of closed-source algorithms; and (5) diminished public trust in rulings perceived as unfair or incomprehensible due to algorithmic involvement. This is followed by real-life examples demonstrating these challenges of AI use in actual courts. Despite these concerns, the study affirms that AI can provide tangible opportunities to reduce judicial backlog, enhance access to justice, and support the training of new judges through interactive simulations of in-creasing complexity. Furthermore, such technologies can facilitate judicial cooperation through the sharing of models and expertise, especially in cross-border or international cases. The study concludes that AI integration in the judiciary must proceed with caution and under full judicial oversight, with the development of clear legal and ethical frameworks to ensure transparency, accountability, and respect for the rule of law.