Hydrogen transportation using existing natural gas pipelines compromises steel integrity because it promotes hydrogen embrittlement and decreases fracture toughness. Accurate and efficient prediction of hydrogen diffusion via steel pipe is critical to ensure the safe operation. This study develops physics-informed neural network (PINN) for modeling hydrogen diffusion in the wall of a gas pipeline. Traditional modeling approaches may struggle with limited experimental data; however, neural networks offer a powerful alternative. PINN was applied to accurately predict hydrogen concentration distributions over time in the pipe wall, despite having a minimal number of experimental data points. The neural network approach was enhanced by incorporating physical laws, specifically Fick’s second law of diffusion. The neural network model achieved accurate results with a mean absolute error of 0.07, using only 6 data points. The results are compared with analytical solutions, and visualizations of the hydrogen distribution are provided. The findings suggest that PINNs are highly effective for this type of physical modeling and promising for further exploration in various physical problems.