A groundbreaking artificial intelligence (AI) tool has been developed to decode brain activity and recreate the images people looked at with remarkable accuracy. The system, called Brain-IT, utilizes functional magnetic resonance imaging (fMRI) data to measure activity in the visual cortex, the part of the brain responsible for processing visual input. By analyzing this data, Brain-IT can predict both the broad layout of an image and high-level information about its content. This information is then passed to an AI image generator, which produces a recreated image that closely resembles the original. The research behind Brain-IT builds on an approach called the Universal Brain Encoder, which starts with an image and predicts the fMRI response that would be expected in individual voxels, small 3D units of brain tissue. This allows the system to learn what each voxel is sensitive to in images, whether it's simple elements like edges or textures, or higher-level abstract categories like faces or scenes. What's remarkable about the Universal Brain Encoder is its flexibility. Unlike traditional image-to-fMRI systems, which are typically custom-trained for each participant and dataset, this model can learn from data from multiple people and transfer information across participants, datasets, or different MRI systems with varying magnetic-field strengths. This breakthrough has significant implications for our understanding of how the brain processes visual information and could potentially lead to new insights into neurological disorders and conditions. The research was presented at the ICLR 2026 conference and has been uploaded to the preprint database arXiv.