The integration of artificial intelligence into the legal system creates new opportunities for optimizing judicial proceedings but is accompanied by serious legal and ethical challenges. The research reveals a fundamental paradox: the most accurate AI systems remain opaque, undermining the fairness of trial proceedings. The study analyzes technical, legal and ethical limitations of AI application, including algorithmic discrimination problems. Based on identified obstacles, practical recommendations are formulated to improve Russian legislation, including the development of a comprehensive federal law on artificial intelligence with a risk-oriented approach.