The expansion of artificial intelligence technologies in the criminal justice system has paved the way for the emergence of new approaches in information analysis and support for judicial decision-making. In the meantime, personalized artificial intelligence, with the ability to process and analyze diverse data and provide an assessment tailored to the characteristics of each case, can play an effective role in stages such as analyzing evidence, assessing risk, predicting the likelihood of recidivism, determining security measures, and proposing appropriate criminal responses. The aim of the present study is to examine the legal and practical capacities and challenges of using personalized artificial intelligence in criminal proceedings. This study was conducted using a descriptive-analytical method and using legal sources and interdisciplinary studies. The findings of the study show that the correct use of this technology can increase the speed and efficiency of proceedings, reduce human errors, make decisions consistent, and improve the proportionality of the criminal response; However, challenges such as algorithmic bias, lack of transparency and explainability of machine decisions, violation of privacy and data security, prediction error and the possibility of undermining the principle of innocence, the right to defense and the independence of the judge pose serious limitations. Therefore, the legitimacy of using personalized AI in criminal justice depends on the establishment of effective mechanisms for human oversight, evaluation and audit of algorithms, control of bias, protection of personal data and the possibility of contesting algorithmic results. The result is that AI should be positioned as a “judicial decision-maker” and not a substitute for the judge, in order to balance technological efficiency with the requirements of a fair trial.