Research and development towards a hydrogen society are actively progressing, leading to an increasing demand for components of various shapes and dimensions. This suggests that, alongside wrought materials, cast and additively manufactured materials may also be utilized for hydrogen-related components, considering machining processes and cost reduction. Additive manufacturing, which offers more processing parameters than casting, enables the optimization of high-performance materials. In this study, additively manufactured 17-4 PH stainless steels with varying porosity levels were produced by adjusting processing parameters, and their hydrogen diffusivities were measured using the desorption method on hydrogen-charged specimens. The relationship between hydrogen diffusivity and porosity was explored using potential models. Three steels, A, B, and C, with porosity ratios of nearly zero (A), 1.5% (B), and 15% (C), calculated from the measured densities via the Archimedes method, were fabricated. A, B, and C were tested in their as-built condition, and their austenite fraction ratios, which significantly affect hydrogen diffusivity, were found to be nearly equal. Despite the differences in porosity, B and C exhibited similar hydrogen diffusivities, both of which were two orders of magnitude higher than that of A. To explain this discrepancy, B and C were modeled as bulk materials with periodic spherical holes based on 2D images. Using Jefferson et al.'s theoretical solution, the calculated diffusivity ratio, assuming the hole diffusivity ranged from zero to infinite, was at most 1.6, which did not align with the experimental results. Another model, assuming isolated grains, was also examined. In this model, hydrogen desorption was considered to occur from individually isolated grains, which were treated as spheres. The diffusion equation for hydrogen desorption from a sphere was applied. Under the average grain size as the sphere diameter, the calculated diffusivity ratio between C and A was around 1,000, showing this model also did not fit the experimental data. Finally, 3D models for finite element analysis were constructed from X-ray computed tomography (CT) images using FUSION360. These revealed that, although the pores appeared isolated in the 2D images, the porosities were substantially interconnected in three dimensions. Diffusion analysis of the 3D models using ANSYS was conducted under the boundary condition that the hydrogen concentration on the surface of the spheres was zero. The calculated diffusivities were nearly equal between B and C, and these values matched the experimental results.