Urine contains significant amount of antibiotics, which must be treated and removed if it is to be used safely as a fertiliser in a nutrient circular economy. The present study systematically studied the efficacy of a granular activated carbon (GAC) adsorption in removing three of the most common antibiotics found in the environment (sulfamethoxazole or SMX, sulfadiazine or SDZ and sulfamethazine or SMZ) from a real nitrified human urine obtained from a pilot-scale urine membrane bioreactor. In addition, the artificial intelligence network procedure with Levenberge-marquate training algorithm was used to model the removal of antibiotics from the nitrified urine. Fixed-bed column tests were performed to acquire breakthrough curves and evaluate the performance of GAC in the adsorption of antibiotics under various operating parameters such as particle sizes (425–1000 µm), adsorbent mass (0.5–1.5 g·L−1), flow rates (0.06–1.8 L·hr−1), adsorption/contact time and pH at 6.2. For all the antibiotics, the maximum antibiotics adsorption capacity was found for a lower particle size, at lower flow rate and higher mass of adsorbent. The breakthrough curves revealed the highest adsorption capacity for SMZ (4.33 mg·g−1) and the lowest adsorption capacity for SDZ (4.01 mg·g−1). Meanwhile, various adsorption models have been employed to evaluate the breakthrough curves of the antibiotics (Thomas, Yoon–Nelson, and Yan models). This study also revealed that artificial neural network can effectively predict (over 99 %) antibiotic removal from a real nitrified urine.